3.1.1. Title of item in the edition: <titleStmt>

The content of element <title type="item">, prominent in the XML file, is the title of that item in our edition and is the one displayed on the project website. It should normally follow one of the following formats: ‘Letter/Note/etc. from […] to […]’ or ‘Diary of […] (startdate – enddate)’. Other titles begin with ‘Journal-letter’, ‘Copy/Copies’, or, less frequently, ‘Transcripts’, ‘Catalogue’, ‘Notes’, ‘Anthology’, ‘Commentary’, ‘Account book’, etc.

NB. Dates are always written out as ‘day [as cardinal number] month [as word in full] year’, e.g. 4 November 1783; see origDate below.

For the participants, namely author and (for letters) recipient, use the name by which they were known at the time of the document (with the exception of Mary Hamilton, who is referred to as such throughout the corpus). If they were previously or later known under a different name or title, please use the following formats:

Different personal title:

Different name:

NB. When confronted with multiple forenames, e.g. Charlotte Margaret Gunning or Martha Carolina Goldsworthy, go with the form most often used by that person when signing as author. So if Charlotte Gunning signs most of her letters ‘Charlotte Margaret Gunning’ (or the equivalent in initials), use that for <title type="item">, rather than ‘Charlotte Gunning’.

If part of a document is missing, the catalogue (mostly ELGAR) will usually open its summary with something like ‘Incomplete note/letter from […] to […]’. Use this formula in <title type="item"> as well (see for example HAM/1/13/10, HAM/1/15/2/7).

The title in <titleStmt> is to be repeated in <msItem> but NOT necessarily in a third title element, namely <collection type="title"> in <msIdentifier>, where it is commented out in the XML. The latter differs in some cases, as it consists of the first line of the summary as provided on the website of the catalogue in question. In probably the majority of items, however, all three titles agree.

3.1.2. Contributions by project team and others: <editor> and <respStmt> in <editionStmt>

The element <respStmt> is retained for research assistants and volunteer or student transcribers, coders, etc. who worked on the file, and for general contributors to archiving, cataloguing and imaging from outside the Unlocking project team. It is no longer needed for members of our team, all of whom are declared differently in <editionStmt> as <principal> or <editor>. Please see <revisionDesc> for entering and dating file actions by members of the team and others.

3.1.3. Acknowledging funders: <funder> in <editionstmt>

The following is to be added to newly transcribed texts [i.e. in the Unlocking project]:

Files dating from the Image to Text project will have one of these entries:

3.1.4. Responsibility for sets of files: <authority> in <publicationStmt>

The following formula is used for newly transcribed texts:

When checking older files that were previously transcribed in the Image to Text project, please make sure that the text is as follows:

3.1.5. Number of words: <measure> in <extent>

The number of words in each item is computed by a script run on the normalised text files, where split words have been reassembled at a line-end. Punctuation is ignored. The word count for each item is then imported back into the XML file. The figures produced by CQPweb may be higher if tokens (e.g. can't as 2 tokens) and/or punctuation are counted.

3.1.7. Title and language in MDC: <msItem>

<msItem> is the element from which MDC takes the title of the item. Make sure that <title> matches <title type="item"> (the element used by the project website). Leave the <textLang> element as shown when the document is in that language.

3.1.8. Number of sheets: <measure> in <supportDesc>

In this <measure> element we count sheets [of paper], like the JRRIL (John Rylands Research Institute and Library), and not pages [of text], like the Royal Archives. According to JRRIL staff, ‘sheets’ is standard practice. For the GEO files, we will have to base ourselves on the images in order to determine how many sheets were used. If the number of sheets is still unclear, the archivist should be consulted if possible.

Make sure there is no period after the number of sheets. Also make sure to have ‘sheet’ rather than ‘sheets’ when there's only one:

For <condition> in the HAM papers, please copy what is provided in the catalogue. If no condition is provided, please word as follows:

Where for the HAM papers we copy <condition> from the catalogue, the GEO catalogue only specifies the physical description, i.e. ‘loose manuscript paper(s)’. After correspondence with the Head of Digital Services, William & Mary Libraries (24/03/2020), this has been put in the element <collation>, and we omit the element <condition>. Future catalogue information from other libraries can go into <condition> and/or <collation> as appropriate.

3.1.9. Hands: <handDesc>

This element in the header is to be written out as follows. This example has 3 hands: Lady Dartrey as the author of the document, the addressee Mary Hamilton as a minor hand (probably), and a third, unknown writer:

The following conventions have been adopted:

• Do not specify whether a document is a letter or note in @scribeRef. Only mention ‘main text’.
• End the content of each <handNote> with a period.
• In the description of the hand, use either ‘author’ (for main hands), ‘writer’ (for someone writing on behalf of the author, see also Documents written on behalf of another above) or ‘writer’ (for minor hands). Do not use ‘scribe’, ‘annotator’, etc.
• Be sure to add the name of the ‘author’ as well as any names of ‘writer’ when known.
• For the name of a hand in @scribeRef, please use the name as given in the first line of the catalogue where the document is from (in this case ‘Lady Dartrey’), not the canonical name we have decided on for the personography, in this case ‘Dawson (née Freame), Philadelphia Hannah, Baroness Cremorne’.
• The optional attribute @cert may be used for major hands that are not entirely certain, with possible values "high | medium | low". Do not use @cert with minor hands.
• When fairly sure of the identity of a minor hand, please enter "addressee | relative | secretary | etc." in @scribe (rather than "unknown"), as well as "psn:xml:id" rather than "psn:unknown" in @scribeRef.
• Please use ‘probably’ (see example), and not ‘possibly/likely/etc.’
• When specifying the location of an annotation by a minor hand on a page, use the preposition ‘at’ instead of ‘in’ for consistency.
• When there is only one minor/major hand, the suffix _1 is unnecessary.

We permit the author to be entered a second time as a minor-hand writer of annotations that are made by them after the original writing (occasionally, as in HAM/2/4, long before) and not considered to be part of the text. See especially Addition vs. annotation below.

NB. When a hand is cross-referenced in the body of the text, remember to add # , e.g.

3.1.10. Date of writing: <origDate> in <fileDesc>

This element contains the date of writing. It should always have the attribute @calendar="Gregorian" and normally also @when, and the date in text, as shown:

If the date is unknown, please include only @calendar and not @when or @when-custom, and put ‘n.d.’ as text:

This is similar to how we treat undated documents in <correspDesc> (see further Overview of dating of letters). As there, it is permissible instead of @when to use @notBefore and/or @notAfter with a <precision> element (see <correspDesc> below). For an item composed over a period, e.g. a diary or a journal-letter, use @from and @to instead. In the latter case, dates of writing in <origDate> will necessarily differ from the single <date> of sending in <correspDesc>:

Note that we treat date of writing of a journal entry as if written on the day concerned, even when a bunch of entries were apparently composed together at a later date.

Note too that the value of @when, @notBefore, @notAfter, @from or @to is normally yyyy-mm-dd. Incomplete dates have the formulas yyyy-mm or yyyy, though –mm-dd (without year) is now disallowed. The attribute @when cannot combine with any of the other attributes discussed here.

When a date is given in text as content of an element like <origDate>, we always use the format ‘day [as cardinal number] month [as word in full] year’, and in that order. There is no -st, -nd, -rd, -th on the date, no comma, nor do we add a leading zero for dates 1-9:

3.1.11. Responsibility for editing: <projectDesc>

For newly transcribed texts [i.e. in the Unlocking project, not Image to Text] in the GEO papers, word as follows:

For newly transcribed texts, word as follows:

3.1.12. Sending and receipt of letters: <correspDesc>

When the date of receipt is noted in the manuscript (characteristic of MH as recipient), a <date> element should be added to <correspAction type="received">. And if the only explicit date in the letter is the date of receipt (as with many of the GEO letters sent by GPW), the <date> in <correspAction type="sent"> should have the attributes @when OR @notBefore and/or @notAfter, as well as a <precision> element (within <date>!). For example:

On the different formats for a date as attribute value and as element content, please see <origDate> above. On the different dates that may be recorded for a letter, see further Overview of dating of letters below.

For the <placeName> elements, fill in the place-name mentioned in the address or dateline of the letter. If only a house or other similar habitation is mentioned (e.g. St. James, The Queen’s Lodge), use the location of the habitation as it was at the time of writing, e.g. London for St. James, Windsor for the Queen’s Lodge. Historic hamlets, e.g. Cowslip Green, are to be considered as places in their own right and so are also tagged <placeName>. Be mindful of places such as Kew that are now part of London but weren’t in the late 18th and early 19th centuries.

If no place-name is mentioned anywhere in the letter/note or address, please use <placeName>unknown</placeName> in both the sent and received entries for <correspDesc>, unless one can be safely inferred, in which case the attribute @cert is added:

For personal names in <correspDesc>, we go with that used by the person in question in their earliest piece of correspondence to Mary Hamilton. For example:

and not ‘Dawson/Freame, Philadelphia, Lady’, since Lady Dartrey did not sign her earlier letters to Mary Hamilton as such.

Please consult the personography. Enter the name as follows:

Obviously the optional née parenthesis applies only to a married woman, but a version with masculine né can be used for a man who changes his surname, usually on adoption or on acquiring an inheritance. Only the title(s) used in the canonical form need be included.

Very occasionally, the optional attribute @cert may be added for a personography identification that is uncertain, with possible values "high | medium | low".

3.1.13. File history: <revisionDesc>

Please enter actions in <revisionDesc> as in the following example, in descending order of date:

NB. Remember to include # before an editor's or contributor's xml:id. Keep the default text, for consistency.

For newly transcribed files, the default text is

We specify year, month and day. If the transcription is done over a period of days (or longer!), we only enter the date of completion.

We use <respStmt> in <editionStmt> to record transcribing or other activities provided by others (see <respStmt> above). Date of completion of their contribution should be recorded in a <change> statement here in <revisionDesc>, though not usually for regular archivists, etc., whose contribution to a specific file is not easily dated.

3.2.1. Page begins: <pb>

All transcriptions with any text at all have a <pb n="1">. The element <pb> is always preceded by a blank line (as a visual aid in the XML file itself — there is no effect on the displayed transcriptions). It is immediately preceded by <lb/>, except in the case of <pb n="1"/>. It can be followed by <cb> on the same line but nothing else other than a footnote, <note resp=…> (so please put any <note hand=…> on the following line with its own <lb/>). The form is always <pb n="nn"/>, where nn = 1, 2, 3, etc. Sequential page numbering is not handled automatically but is the editor's responsibilty. After the first <pb>, an element such as <opener> or <p> must be opened to contain text. See also Divisions below.

3.2.2. Column begins: <cb>

If a single image displays two pages, as is frequently the case for HAM letters from the JRRIL, do not code as separate pages:

Instead code as two columns:

Don't repeat <pb> before the second <cb>. When <cb> is the first element in a line, necessarily therefore at least n="2", it must be preceded by <lb/>, and again before that, a blank line. Such blank lines have no effect in XML but they help an editor to quickly relate the transcription to the image.

Although use of ‘column’ is a bit misleading, because for the author they were separate pages folded back to back, we will continue using <cb> for the JRRIL letters (advice in correspondence with the JRRIL Imaging Manager and TEI Project Co-ordinator). It does mean, however, that each <pb> corresponds to one image.

3.2.3. Line begins: <lb>

The element <lb> can have an optional @rend that applies to the following line rather than a whole element (see also Overview of attribute @rend below), with possible values "indent", "center" or "align-right" (but not "inline", which would be redundant at line beginning). With the present XSLT display script, <lb rend="center|align-right"/> does not create an extra blank line after itself, which can be useful, as other elements with such a @rend attribute create a blank line afterwards and also force any items on the same XML line but outside the @rend element onto a separate line in the display.

Represent wide blank vertical spaces in the original with an extra empty <lb/>, e.g. in HAM/1/8/2/22:

3.2.4. Paragraphs: <p>

After the opening paragraph, the element <p> is used for new paragraphs as visually indicated by the author of the text, e.g. by an indented new line, a blank line inbetween two other lines, a previous line that ends well short of the right margin, or an inline mark (e.g. GEO/ADD/3/82/4, end of p.1). Occasionally there is a longer than usual space before a semantic paragraph that starts in midline (e.g. HAM/1/2/31, HAM/1/15/2/3). Such ‘new paragraphs’ that continue on the same line need <p rend="inline"> to prevent the display script from moving to a new line:

The author's desire for a new para has clearly been trumped by the need to save paper. Our website retains the author's layout. We don’t mark a new paragraph inferred only from content and without any visual sign.

In the header, <p> never has an attribute. In the body it may have the attribute @rend (see Overview of attribute @rend), with values "align-right" or "center" (which shift the whole paragraph), or "indent" (which indents the first line only). As noted for the example just above, the value "inline" is needed in some cases; see also the second example in Postscripts below and the fuller explanation in Overview of attribute @rend.

As for the end of a paragraph, we routinely have to end <p> in midline before <closer>, because the current TEI guidelines do not allow <closer> within <p> (cf. Forney et al.’s paper 2020). To avoid a line-break in website displays after the closing </p> tag, rend="inline" must be added to the opening <closer> tag that follows:

3.2.5. Divisions: <div>

The element <div> is a subdivision of <body> (at least in our project, since we do not use <front> or <back>). It is needed in one or more of three different situations, discussed in separate subsections below.

Note that if an item contains a <div>, the whole of the <body> must be distributed between one or more <div>s, with no residue (such as a footnote or annotation) orphaned outside any <div>.

Note also that if any whole <div> is written by a different hand from any other, then all the <div>s in the item must have the attribute @hand. Furthermore, as div/@hand does not show on-screen, a footnote after an opening <div> tag should explain the change of hand.

Possible values for the obligatory attribute @type are:

• "letter" (for the first <div> of an item that the catalogue describes as a letter, or of an embedded letter)
• "note" (for an item that the catalogue describes as a note)
• "continuation" (when an author indicates resumption of letter-writing)
• "entry" (for a diary/journal entry)
• "address" (for the ‘direction’, which may be just a name, or the postal address in a letter or note)
• "cutting" (for a newspaper cutting)
• "part" (generic type for a <div> that doesn't fit any of the others)

Don't leave a blank line between a closing </div> and the next opening <div>: the 8-space indent is enough to make the div change conspicuous in the XML.

NB. The first of the two examples above is likely to be problematic in MDC without some recoding, as it involves a midline div and para break without rend="inline" on the second para.

If a div break coincides with a page break, please make sure that that the div break comes first, so that <pb> and its customary preceding blank line occur at the start of the new <div>:

3.3.1. Openers: <opener>

The opener ‘groups together dateline, byline, salutation, and similar phrases appearing as a preliminary group at the start of a division, especially of a letter’ (TEI Guidelines). In this edition, <opener> can include the following elements: <dateline>, <date> (inside <dateline>), <placeName> (also inside <dateline>), and <salute n="opening">.

NB. <opener> refers to the time of production (TEI Guidelines). Therefore metadata to do with the date received rather than date sent, as is typical of Hamilton’s annotations, should appear outside <opener>. (On <date> in this context, with or without a <dateline> wrapping, see Dateline.)

A salutation such as ‘My dear X’ at the very start of the letter is tagged with <salute> inside <opener>.

NB. If a letter doesn't begin with a salutation, the first sentence belongs to <p> rather than <opener>. Since <salute> is disallowed in <p>, a salutation later in that sentence, as in e.g. ‘I have so missed you my dear X, […]’, is instead marked with the project-specific tag <seg type="openSalute"> (see Salutations and Overview of person references).

Each <div> of a multi-div document can in principle have its own opener, but when a <div type="continuation"> is used, <opener> is not repeated (as already noted under Diaries and journal-letters above). If a weekday is mentioned to indicate when the letter was continued, use only <dateline> (containing <date> if a date is given or can be inferred). That <dateline> element leads straight into <p>.

<opener> has an optional attribute @rend for rightwards shifting of the content, see also Overview of @rend.

3.3.2. Closers: <closer>

A <p> element needs to be closed before a <closer> is opened; see also Paragraphs.

The closer ‘groups together salutations, datelines, and similar phrases appearing as a final group at the end of a division, especially of a letter’ (TEI Guidelines). In this project, all text in <closer> should be inside one of the elements <salute n="closing">, <seg type="closeSalute">, <seg type="closeFormula">, <dateline>, <note> or <signed>, leaving no untagged text.

If the text leading up to the closer includes words and phrases such as ‘adieu’, ‘God bless you’, ‘Give my love to X’, ‘remember me to X’, etc., they should all be tagged as part of the preceding <p> (but see also <salute n="closing">).

Since <closer> (like <opener>) refers to the time of production, an annotation to do with the time of receiving the letter should sit outside.

It is possible to have more than one <closer> in a document, e.g. one in the main body and a second one

• inside a <postscript> (see Moved text for an example)
• after a <postscript> (e.g. needed to wrap a dateline there)

<closer> has an optional attribute @rend for rightwards shifting of the content, see also Overview of @rend.

3.3.3. Salutations in opener and closer: <salute>

Salutations concern references to the addressee. Those in <opener> or <closer> are enclosed in <salute> with the attribute @n, value "opening" or "closing" as appropriate.

NB. For salutations inside <p>, where <salute> is disallowed, see the use of project-specific tags in Segments and Overview of person references.

The attribute @rend does not need to be specified in <salute> (this can be done in <opener>, <closer>, or indeed <lb>; cf. Overview of @rend).

3.3.4. Datelines: <dateline>

Only use this element for dates and/or place names written by the author at the start or end of a letter or of a <div>. Include it as part of either <opener> (except in a continuation <div>, see Openers above) or <closer>:

NB <dateline> is also to be used in the absence of <date> and <placeName> elements if information regarding the creation of the letter is still present, e.g. when only the day of the week is mentioned.

If a <dateline> is present for authorial text, any annotation by a later (minor) hand that adds to or corrects the date or place of sending goes inside it. See e.g. HAM/1/5/2/2:

But if all the metadata concerning a date or place-name of sending is written by a minor hand, there will be no <dateline>. The annotation(s) will contain a <date> and/or <placeName> and will not be part of <opener> or <closer>. See e.g. GEO/ADD/3/82/10:

If <dateline> appears at the end of a letter, it should be included in the <closer>, normally after <salute> and <signed>. If there is a <postscript> intervening between <signed> and <dateline>, wrap <dateline> in a second <closer> element:

3.3.5. Dates: <date>

This element is to be used for the date only, and not for other parts of the dateline or annotation such as time of day (e.g. ‘8 o’clock’) or the name of the day (e.g. ‘Tuesday’). If the date is split by ‘received’ or a time, use separate <date> tags for the fragments. The name of the day can be left inside <date> if adjacent to the numbers but doesn’t require a <date> element if not.

An exception has been made in diaries for <div>s that would otherwise contain no date, if the day can be dated. And in other cases, a date without textual content has been inserted — such dates should have an @resp to indicate the editor responsible for the inferred date. Both uses of <date> are to ensure that persons mentioned in diaries are only counted once per day for the purposes of social network analysis.

Typically, the month and day fragment will receive a full @when value ("yyyy-mm-dd"), while the year fragment just gets the year ("yyyy"). If the year was clearly added after the creation of the letter (from evidence of continuity, size, position or ink), then it should be placed in a separate <note[@hand]> inside <date>.

NB. When a date is given as ‘The […] of […] etc.’, please include ‘The’ inside <date>.

NB. In our project, <date> has paired opening and closing tags around textual content, so the following attribute-only structure is not used: <date when="[…]"/>.

3.3.6. Postscripts: <postscript>

TEI requires text inside <postscript> to be embedded in <p>, e.g.:

Either <postscript> or the preceding <lb> can have an optional @rend. If there is a <label> for explicit use by author of ‘P.S.’ or ‘Postscriptum’, etc. and the text of the postscript continues on the same line, <p> requires rend="inline", whether or not there is an earlier @rend in <lb> or <postscript>, e.g. GEO/ADD/3/82/69:

NB. Remember to leave a space between the closing tag </label> and the opening tag <p rend="inline">.

If the writer adds an afterthought to a letter without an explicit ‘P.S.’, use <postscript> regardless, but without <label>.

A postscript normally contains only <p>, but when a writer uses words and phrases that need wrapping in elements normally reserved for <closer> (e.g. <signed>, <salute n="closing">, <seg type="closeFormula">), please add a <closer> element to contain them:

For information on using the elements <anchor> and <ref> with postscripts that need to be moved, see Moved text. If a moved postscript contains elements other than <p>, e.g. <label>, <closer>, <signed>, then <ref> must be repeated inside each such element, as in the example above.

NB. The TEI Guidelines disallow <postscript> as first element in <body> or <div>. If a fragmentary letter consists solely of a postscript, we precede it with a blank line containing an empty <p/>.

3.4.1. Names

The elements <persName>, <placeName> and <rs> are used to tag names.

Personal names are stored in a personography and indexed by means of an xml:id unique to each individual. Place-names, though tagged, are not similarly indexed in this project. Spellings of personal names are not normalised to a standard form, nor most place-names (but see Spelling in names below), so the differences between diplomatic and normalised transcriptions will concern abbreviated and suppressed names.

An abbreviated name should be tagged with <abbr>. If the full form is known, it is marked up in a <choice> element, as detailed in Expansion of abbreviations below:

If a name is partially suppressed by means of a horizontal line, use two em-dashes for the line and surround the whole form with <abbr>:

The suppressed name, like an abbreviated one, should be expanded if known:

If a name is represented solely as a horizontal line, use two em-dashes wrapped in <abbr>, even though complete suppression of a name is not, strictly speaking, abbreviation. However, if we know or discover what name has been suppressed, the <abbr> can be wrapped in <choice> and the full form given in an <expan> element. Thus ‘a M^r– —— who's name I dont at this moment recollect’ would be coded:

NB. Do not include final punctuation in a name element. Characters under a superscript in an abbreviation are obviously not excluded.

3.4.2. Passages marked for research analysis: <seg>

The element <seg> in TEI is an arbitrary segment, a general-purpose tag for stretches of text. In this project, its attribute @type can carry one of seven values: "para" | "rp" | "openSalute" | "bodySalute" | "closeSalute" | "closeFormula" | "author". All but the first are used to mark up text for analysis in the research strands of the project.

• type="para" inside <summary> in the TEI header represents a paragraph.
  • This simply allows paragraph breaks in the library catalogue to be taken over into our metadata (following TEI Project Co-ordinator 16/1/20).
• type="rp" marks passages relevant to reading practices.
  • See also Quotations below.
  • The reading practices research strand investigates the reading, writing and circulation of texts among participants in the Papers, including printed and manuscript material, private letters and other textual media. For the purposes of this research strand, ‘prints’ (i.e. printed images accompanied by textual content, such as engravings) are considered texts.
  • NB. <seg> cannot cross a paragraph <p> boundary, so a long passage relevant to reading practices must be broken into separate <seg>s if it extends beyond the end of a paragraph.
• type="openSalute | bodySalute | closeSalute" marks various kinds of reference to the addressee.
  • They belong in <opener>, <p> and <closer>, respectively.
  • These and the remaining values of @type are part of the mark-up used in the research strand on sociolinguistics and letter-writing to capture non-structural features that lack a standard tag in default TEI mark-up.
  • See Salutations and Overview of person references.
• type="closeFormula" marks formulaic references to the author in <closer>, such as ‘yours (most) affectionately’, ‘ever your most humble servant/friend’, etc.
  • See Referring to the author for more detail on this segment type.
• type="author" marks the occasional self-reference to the author of a text, e.g. when John Dickenson refers to himself as ‘your dear Husband’.
  • This kind of self-reference is generally found in <p>, though there are a handful of examples in <closer>.

All references to persons — and there is always one at least — inside the <seg> types "openSalute | bodySalute | closeSalute" (as well as <salute n="opening">) require a <persName> or <rs> tag, the @ref attribute pointing to an addressee's xml:id. Likewise type "author", in this case with @ref pointing to an author's xml:id. Final punctuation goes inside the <seg> or <salute> element, but outside <persName> | <rs>.

NB. If <foreign> and/or <emph> are needed inside <seg>, the preferred relative order from outside to inside would be <seg><rs><foreign><emph>.

3.4.3. Direct speech: <q>

Use <q> to represent direct speech by a real person, as when a letter-writer quotes someone they met and spoke to. If there are quotation marks in the original, reproduce them inside the <q> element. Indirect speech may also be signalled by <q> if the writer uses quotation marks.

3.4.4. Quotations: <quote>

Use <quote> where a writer quotes from another letter or diary, and reproduce their (non)use of quotation marks inside the <quote> element (see Quotation marks below). The passage containing the quotation should be tagged with <seg type="rp">.

NB. We use <quote> (with final -e) to wrap and thereby identify a passage of quotation. We do not use the entity " to represent the quotation mark character.

The passage containing such a quotation should obviously be tagged with <seg type="rp"> as well. By default, <seg> and <cit> are co-extensive (<seg> wrapping <cit>), but <seg> can be extended to left (as in the example above) and/or right of the citation if there is adjacent text that belongs inside it.

The bibliographic citation is not part of the author's text and only appears as a tooltip when the quotation is moused over. For bibliographic citations that do appear in the text display, almost always in editorial footnotes, see Bibliographic citations below.

3.4.5. Verse line: <l>

The tag serves to distinguish verse from prose, and it exempts the part-of-speech tagger from operating where register, style, spelling and often date can be quite different from the surrounding context. There is no need to normalise in verse lines.

3.4.6. Foreign words and text: <foreign>

Languages other than English are tagged <foreign>, with the attribute @xml:lang set to the appropriate language, most notably French. Common values are "fr" | "it" | "la". We have needed "pt" [Portuguese] and "nl" a handful of times. To determine what counts as French and what English at the time of writing, please first consult the list of French words and phrases in the GH repository and follow that. If the word/phrase is not mentioned in that list, check the OED and consider the context of the letter, as well as the frequency with which this particular word/phrase occurs in all transcriptions up until this point. After this, a suggested classification (i.e. either English or French) should be put to the team before being entered in the list of French words and phrases in GH, so that future instances can be tagged — or not — consistently.

When a word or phrase originally borrowed from French is considered to be part of English at the time of writing (e.g. Adieu), leave it unmarked for <foreign>. However, when used in a French context, include in <foreign> with the rest of the line/text in French:

3.4.7. Indications by author for recipient: <fw>

The element <fw> is normally used in TEI for representing running headers, catchwords, etc. in printed books. We use it for authorial annotations for the benefit of a contemporary reader, especially the recipient of a letter, such as catchwords, headers explaining what follows, pagination, stanza numbers, and ‘PTO’ instructions. They should not form part of the linguistic corpus in the normalised text. Some examples follow:

• True catchword inserted below the line at bottom right (HAM/1/4/4/1):
  <lb rend="align-right"/><fw type="catch" place="bot-right">Country</fw>
• Probable catchword at end of last line on page (HAM/1/4/4/9):
  <lb/>[…] for this <fw type="catch" place="inline">picture</fw>

  NB. Such inline cases require a decision: was it an intentional catchword, or was the repetition overleaf a mistake? — in which case tag the second occurrence as <surplus> instead of the first as <fw>.

• Where the author has numbered the pages or otherwise indicated order of reading, or given some other explanatory heading (GEO/ADD/3/82/14, 24 and 69):
  <lb rend="center"/><fw type="pageNum" place="top-center">2<hi rend="superscript">d.</hi>.</fw> <lb rend="center"/><fw type="head" place="top-center">Continue on P.S.</fw> <lb/><opener><dateline><fw type="head" place="top-left">Tuesday Night in answer to Monday Night.</fw></dateline>
• Where the author has numbered the stanzas of a poem (GEO/ADD/3/82/37):
  <lb rend="center"/><fw type="stanzaNum" place="center">2<hi rend="superscript">d.</hi>.</fw>
• Where the author has given an equivalent of modern P.T.O. (HAM/1/4/7/5):
  <lb rend="align-right"/><fw type="turnPrompt" place="bot-right">Turn</fw>

NB. The attribute @place in <fw> is purely for information in our mark-up and has no effect on display. Therefore any left-right positioning other than left-aligned or inline requires a @rend attribute in <lb>.

3.5.1. Underlining: <emph>

By default, underlining is tagged as <emph rend="underlined">.

When several words are underlined individually, each should be tagged separately, whereas only one tag is necessary if the same continuous line underlines several words. Compare the examples from GEO/83/82/11 and GEO/3/82/13 below:

If underlining extends over two lines, use separate tags for each line. For example, in GEO/3/82/43:

Leave any final punctuation outside the tag, but a continuously underlined passage may include non-final punctuation.

NB. Use more specific values for @rend where appropriate ("double-underlined" | "dotted-underlined " | etc.) Although the first part of the value is ignored at present, the last part will still trigger single underlining in our display script, and the more precise mark-up may be helpful to a future researcher.

We have (almost) stopped using <hi rend="underlined">, retaining it in principle only for underlining that couldn't possibly be emphatic. Even if a writer such as CMG consistently underlines foreign words, we can't rule out some degree of emphasis, so @rend goes in <emph> rather than being an attribute of <foreign> (as formerly in the project) or <hi>. Our default in such cases is <foreign> wrapping <emph>, but we do have examples the other way round (46 to 322).

NB. Underlining occasionally gives a clue as to whether a <foreign> tag is needed, as in this extract from HAM/1/15/1/17, where lecture in the fourth line is clearly not the English word but rather the French lecture ‘reading’, while in the second line, underlining indirectly supports the transcription Drame rather than Drama:

3.5.2. Superscript, subscript, italics, etc.: <hi>

The element <hi> (highlighted) ‘marks a word or phrase as graphically distinct from the surrounding text, for reasons concerning which no claim is made’ (TEI Guidelines). The attribute @rend is obligatory. Use the value "superscript" for characters at the end of an abbreviation that are raised above the baseline. The display script presents them in a smaller and slightly raised font. Much less frequently needed is "subscript".

NB. Superscript is to be distinguished from insertions above the line, which are tagged as <add place="above"> and displayed wholly above x-height; similarly subscript vs. <add place="below">. See Additions.

Use <hi rend="italic"> for linguistic forms, book titles, etc. in footnotes. NB. It is a minor, though not uncommon, tag abuse to use @rend in editorial footnotes rather than to indicate ‘how the element in question was rendered or presented in the source text’ (TEI Guidelines).

Some other available values for @rend such as "larger" | "retraced" are there for completeness of transcription, but if used are not visible in the display.

3.5.3. Hyphens, dashes and lines for suppression

For hyphen (a word-internal feature), use a single hyphen in these contexts:

• Normal hyphen (well-known), whether intra-word on one line or at the end of a line where the word has been split by the author across two lines.

  NB. Only use hyphen at end of line (no following space) if the author does so; sometimes they use a double hyphen (=) or nothing at all.

  NB. The hyphen may also be repeated before the word fragment at the start of the next line, but contemporary practice is rather variable.

• Horizontal mark under superscript abbreviation (M^rs–)

NB. A single hyphen never has a space to its left.

For dash (a piece of word-external sentence punctuation), use two hyphens with a space on either side of the pair, even at the beginning or end of a line:

A horizontal line used to suppress all or part of a name or an oath is represented by two em-dashes not surrounded by spaces. This is not sentence punctuation, merely suppression (‘blanking’), and our transcription is a visual representation of the manuscript form.

• Wrap the suppressed word in <abbr>:
  Whilst I was at <persName>M<hi rend="superscript">r</hi> <abbr>——</abbr></persName> […]
• In turn, if the unsuppressed form is known, wrap <abbr> in <choice> with <expan>:
  <seg type="bodySalute"><rs> you little <choice><abbr>D——l</abbr><expan>Devil</expan></choice></rs></seg>
  the <persName><choice><abbr>K——g</abbr><expan>King</expan></choice></persName> &amp; <rs>my <choice><abbr>B——r</abbr><expan>Brother</expan></choice></rs>
  <seg type="author">her <persName><choice><abbr>D——</abbr><expan>Dickenson</expan></choice></persName></seg>

  Whether complete suppression of a word can fairly be called abbreviation is open to question, but the use of <abbr> even here allows for expansion if the unsuppressed form is known, as well as serving to distinguish further a line of suppression (which if not tagged would often have been ‘ —— ’) from a dash (‘ -- ’): it's important to keep them apart.

3.5.4. Authorial lines and flourishes: <milestone>

We use the <milestone> element to represent some kind of line or flourish added by author or annotator to mark the conclusion of one section, e.g. underneath the dateline or a postscript (to the extent that this can be distinguished from word underlining or dashes). If a simple forward slash ‘/’ or other character is a better representation of such a mark, and when it is inline, use that instead (it’s about the visual representation, not the function of the element <milestone>).

The attributes and their possible values are as follows:

• value of @style: "20% align-left | 20% center | 20% align-right | 50% center | 100%".
  • A crude indication of width and position.
• value of @type: "line".
  • Only "line" at present, even when the actual mark is diagonal, or distinctly bowed, or with curlicues.
• value of @unit: "address" | annotation | date | dateline | entry | letter_body | pageNum | paragraph | postscript | salutation | section | signature | time".
  • The part of the document that is closed off by the line or flourish.

When a <milestone> is used to mark off a full element (e.g. opener, salute, dateline, note, etc.), place it inside that element, before the closing tag:

The annotation in this example has been rounded off with a flourish. By ensuring that the milestone is marked up as part of the annotation, the line will — as it should — disappear from the normalised display.

3.5.5. Quotation marks

Curly quotation marks are available in oXygen from the Symbols menu (Ω button on toolbar and search for quotation in the character map). Recently used symbols remain in the menu for quick re-use.

3.5.6. Apostrophes

We distinguish apostrophes from single quotation marks. Always use the straight apostrophe ('), both in contexts where apostrophe is still used (possessives, shortened auxiliary verbs, etc.) and in contexts no longer standard:

NB. The last two examples here would of course be normalised to <reg>distressed</reg> and <reg>would</reg>, respectively.

3.5.7. Parentheses

When a pair of characters is used to surround a parenthesis, always use (…) even if both opening and closing brackets look more like straight /…/.

When a bracket-like character seems to be used singly as some sort of punctuation mark (and not merely careless omission of the closing bracket of a parenthesis), code it with whatever character comes closest to it visually, whether /, ( or [.

3.5.8. Ampersands, etc.

A variety of single-character squiggles are used by writers to abbreviate the conjunction and. We represent any of them by an ampersand. However, the character ‘&’ has a special value in XML and must be entered as the XML entity ‘&amp;’ in the text stream.

Other internal entities include angle brackets (aka the less-than and greater-than symbols), used all the time in mark-up but rarely needed in text. If needed, they are to be coded as ‘&lt;’ and ‘&gt;’, respectively. As for straight apostrophe, we use the normal keyboard character ' rather than the entity ‘&apos;’ — apparently without ill effects.

3.5.9. Word spacing

Spacing can be idiosyncratic, and we often give writers the benefit of the doubt or even silently modernise an apparent missing space, e.g. alittle Friend > a little Friend. However, wrong intra-word spacing, e.g. a lone for ‘alone’, is corrected with <sic> + <corr> to produce alone in the normalised edition, except for instances that would have been considered appropriate at the time (e.g. to day, my self, etc.), in which case we don't intervene.

For the special case of a word broken across a line boundary (i.e. hyphenation, with or without an explicit hyphen), see Word separation at line-end below. For initials and acronyms written without spaces, see Expansion of abbreviations.

3.5.10. Additions: <add>

The element <add> (addition) ‘contains letters, words, or phrases inserted in the source text by an author’ (adapted from TEI Guidelines). The obligatory attribute @place indicates where the text to be added comes from in the original document:

• place="above" or "below" [the line]
• place="margin-left" or "margin-right"

  when the addition is roughly level with the line

• place="inline"

  when spacing, other edits or ink make clear that something was squeezed in on the line as an afterthought, or as part of a substitution

• place="center | top-left | top-center | top-right | bot-left | bot-center | bot-right"

  when the addition comes from a location distant from the line where it is to be inserted

The values "above" and "below" of @place affect the display visibly, while all other values suggest that the added text can be inserted at the default level of the line.

We no longer use the optional attribute @hand.

3.5.11. Deletions: <del>

This element is used for text (anything from one character to a long passage) deleted by the author of the document, or otherwise indicated by them as being superfluous or spurious. The deletion is visible in the original.

NB. Not to be used for deletions made by annotators, editors or encoders; see further below.

The obligatory @rend attribute most often has the value "cancelled" (i.e. crossed out), "erased" or "overwritten". You can use up to two values together, including the fourth possible value, "censored". This last can only be used in the relatively infrequent event that the author has self-censored themself. It is much more common for an annotator to censor text (dealt with in Censorship below). Because censorship by another is not coded as <del>, <del> does not need an attribute @hand.

3.5.12. Censorship: @reason="censored"

‘Censorship’ as used here is the deliberate obscuration of information, most often personal names, especially by MH herself in the papers she archived, whether in her own diaries, in copies of letters sent, or in letters she received. Sometimes it involves place-names, notably in her obsessive scrubbing-out of ‘(N)orthampton(shire)’ in letters received from William Napier, where the place is hardly a secret. Sometimes it involves whole passages.

When the censor is the original author, even if carrying out the deletions possibly long after the original composition, we usually ignore the passage of time and use <del rend="… censored">.

Where the censor is an annotator, deletions are not coded by <del>, which is an authorial action. Instead they must be coded as either <gap> or <unclear> (both covered below under No transcription and Cautious transcription, respectively), with "censored" included in the values of @reason:

Censorship by an author of their own earlier writing raises the same questions as authorial insertions, discussed in Addition vs annotation above. In those few documents where a set of insertions and deletions by the author can be safely distinguished from the original writing, and where we choose to treat such insertions as annotations by a minor hand (using a separate <handNote> entry in <handDesc>), then the author-censor's later ‘deletions’ are likewise those of an annotator, so <del> is not used.

3.5.13. Change of hand: <handShift>

The element <handShift> is used sparingly. It has the obligatory attribute @new, whose value indicates the hand and only indirectly the writer. It is only used when multiple hands have contributed to the same (section of) text, e.g. when a signatory does not write the rest of the letter, or when several people contribute short passages to the same letter.

However, if the contributions of different hands can be neatly divided among separate letters within the same document, or at least into cleanly demarcated sections such as letter content vs. address page, use <div> with the attribute @hand rather than <handShift>, e.g.

The display scripts do not show a change of hand, whether coded by either <handShift> or <div>, so it is helpful to the reader to insert a footnote after the <pb> tag if the change of hand coincides with a page break, or after the <handShift> or <div> tag if mid-page:

If a hand adds anything after the letter was sent, that is an annotation and clearly distinguished on-screen (see just below), so neither a <handShift> tag nor a footnote is needed.

3.5.14. Annotations on manuscript: <note>[@hand]

We use the term annotation for anything written on a letter after its sending, including notes by the recipient, by members of the Hamilton/Anson family curating the collection, and by archivists employed in a library or archive. Anything written after sending by a different hand from the author(s) of the letter must be an annotation.

NB. Although franked addresses are written by another hand at a different time from the letter itself, they must at least be written before sending and so are not classed as annotations. See Frank signatories above.

With drafts or copies of out-letters, the position is less clear, and with diaries there is no ‘sending’ (though in fact MH often reports sending portions of her diary to friends). Text in hands other than the (current) main hand must be an annotation, but text in the main (author's) hand may be an addition or an annotation. For considerations we can bring to bear on the choice, see Addition vs. annotation above.

An annotation is coded as <note> with the attribute @hand, preferably given as the first attribute for visual checking purposes, to distinguish an annotation from a footnote, which is <note> with attribute @resp. (Order of attributes is actually immaterial in TEI/XML.) The attribute @place indicates a rough placement on the page but has no effect on the display. An optional attribute @rend can be used to adjust L-R placement on the line, though increasingly such a @rend attribute is placed in <lb> instead (see Overview of @rend below).

The display scripts display annotations in the diplomatic transcription in a different font and colour from those of the main text, with a mouseover tooltip attributing the hand, if known. Annotations in letters that concern the letter as a whole, e.g. archivists' numbers or shelfmarks, are often moved outside the main content if not already at the very top. Annotations are not considered part of the textual corpus and are omitted completely from normalised transcriptions.

If an annotation appears above the line and is to be displayed accordingly, the only mechanism we have for raising text above the line is <add place="above">, and so we have occasionally used <add> on the whole contents of <note[@hand]>; strictly speaking, that is not an addition:

3.6.1. Diplomatic vs. normalised: <choice>

A <choice> element always wraps a pair of elements together, with no residue. We use three different pairs, discussed below. The first member is always the one selected for diplomatic outputs, the second for normalised.

NB. There are circumstances where (usually) the second element itself contains an embedded <choice> (see Self-embedded <choice> below).

NB. We don't normalise quoted text, poetry or foreign-language text.

3.6.2. No transcription: <gap>

The element <gap> ‘indicates a point where material has been omitted in a transcription […] because the material is illegible [or] invisible’ (adapted from definition in TEI Guidelines). It signals an editorial decision to omit, or an inability to transcribe, a span of text that follows that point. It does not itself extend over a span, so the closing slash / goes at the end of the tag (as with <lb/>) and there is no separate closing tag:

As indicated by these examples, <gap> must contain the attributes @reason and @extent:

• value of @reason: up to 3 from "cancelled | censored | covered | crease | damage | end-of-line | erased | faded | fold | folded-corner | illegible | ink-blot | overinked-pen | page-cut | seal | smudge | tear | unclear".
• value of @extent: "nn character(s) | word(s) | line(s)", where nn = 1 | 2 | 3 | etc.

  The value "½ line" is also permissible.

  NB. A hedge like "5-6 characters", "2-3 words" is permissible, and the display script reacts to the first number only.

If an untranscribed passage crosses line boundaries, it should be tagged as a single <gap>:

When an author has written something and subsequently deleted it, rendering it irrecoverable, <del> and <gap> should be used together as follows:

The method of deletion is stated in @rend for <del>, e.g. "cancelled" or "erased", so it need not be repeated in @reason for <gap>: a sufficient reason for a gap in deleted text is that it is "unclear" or wholly "illegible".

NB. If a part of the text has been censored by the recipient and is no longer legible, do not use <del> with <gap> but <gap> alone.

The <gap> element should only be used where text has not been transcribed. If partially legible text has been transcribed, <unclear> should be used instead.

3.6.3. Cautious transcription: <unclear>

If text is legible with a reasonable degree of certainty, use <unclear>:

Permitted values for the obligatory attribute @reason in <unclear> are "cancelled | censored | corrected | covered | crease | damage | end-of-line | erased | faded | fold | folded-corner | indistinct | ink-blot | overinked-pen | overwritten | page-cut | seal | smudge | tear". The default value if there is no specific cause for the unclearness is "indistinct". There is no limit to how many values can be used together:

3.6.4. Supplied text: <supplied>

The tag <supplied> is used as sparingly as possible. Words omitted by an author in error are not <supplied> in the transcriptions; a footnote may be offered instead. Where damage leaves no visible trace of a word, <gap> is used. If a word or character is almost readable, but with some uncertainty, it is marked <unclear>.

The tag <supplied> appears in this edition when there is not enough text remaining to justify <unclear>, but where there are sufficient traces of text to corroborate contextual and physical evidence of loss. This evidence may include third-party writings, such as a recipient who adds the words that were once there but now gone due to damage (see HAM/1/2/20). An example:

As indicated in the example, there are three obligatory attributes, @resp, @reason and @cert:

• @resp as usual indicates who is responsible for the editorial action.

  NB. Remember to include # in the value.

• value of @reason: up to 3 from "covered | crease | damage | end-of-line | faded | fold | folded-corner | illegible | ink-blot | missing | page-cut | seal | smudge | tear".
• value of @cert: "low | medium | high", with default "high", since <supplied> is not to be used lightly.

  A lost possessive s or 's may be <supplied>, but normally with cert="low", since the use of apostrophes at that period is rather variable. Likewise a lost past tense or past participle (e)d may be <supplied>, but normally with cert="low" or cert="mid", depending on the consistency of that author's selection from among such forms as wished, wish'd, wishd, wisht (all of which occur in the corpus).

Very rarely indeed we have wrapped <supplied> inside <add> in order to have the supplied text placed above the line (HAM/1/2/47):

3.6.5. Redundant words: <surplus>

When a word is mistakenly repeated by an author, e.g. ‘There is a walk near the the Castle called […]’, tag the second occurrence as <surplus> with the attribute @reason and value "repeated". Other possible values for @reason in <surplus> are "anacoluthon" | "inappropriate" | "inserted-material", the last-named for the situation where a change of mind by the author in another part of the sentence makes the tagged word unnecessary in the revised version.

If a word is repeated over a page break, there is the possibility of tagging the first one as a catchword instead, if that would seem consonant with the writer's habits.

3.6.6. Moved text: <anchor> and <ref>

An <anchor> ‘attaches an identifier to a point within a text, whether or not it corresponds with a textual element’ (TEI guidelines). A <ref> (reference) ‘defines a reference to another location […]’. When we move a segment of text to a more logical reading position in order to avoid a discontinuity, we link source and destination positions with an <anchor> and a <ref> element, respectively. The chunks of text liable to be moved are addresses, annotations, dates, postscripts and ‘sections’ (a generic term for anything from a few words to several paragraphs).

Suppose a postscript has been written at the very top of the letter (see HAM/1/10/1/25, GEO/ADD/3/82/11 for examples). In the transcription, an <anchor> is placed at the top, marking the original position of the postscript and displaying as ▼. The element is a point, not a span, so the closing / is contained within the tag. The postscript is moved to where it would normally have been located had there been space enough, namely the foot of the letter. The moved text is wrapped in a <ref> whose attribute @target, "#ps", corresponds to the @xml:id of the <anchor>, "ps". However, the attribute @type on <ref> with value "moved" instructs the display script not to create a hyperlink tempting the reader to jump to the anchor position, which would be the more normal use of <ref>[@target] (on which see Links and cross-references below).

Top of the letter (place of the postscript in the ms):

Bottom of the letter (place of the postscript in the transcription):

For the placement of <ref> inside the postscript, see Postscripts above. Recall that if there is a <label> before the <p> of the postscript, <ref> must be duplicated inside <label>:

Anchor requires three attributes: @xml:id, @type and @resp:

• value of @xml:id: "ps | ann | sec | add | dt".

  The one actually used in a particular <anchor> must correspond to the full form given as the value of @type.

• value of @type: "postscript | annotation | section | address | date".

  The schema prompts with possible values and ensures that @xml:id and @type belong together.

  If there is more than one of any <anchor> type in the document, the values of their @xml:id's are numbered, e.g. "ps1 | ps2 | …", "sec1 | sec2 | …".

• @resp as usual indicates who is responsible for the editorial action.

An <anchor> should not be free-floating but should have a <lb> to its left. Remember to include @resp in <anchor>, with # before your initials.

Although the original location is displayed as ▼, the moved text with its <ref> wrapping has no special marking on-screen. We generally add a footnote at the end of the moved text to alert the reader, often with this formulaic wording:

The parts about unfolding and direction of writing are each optional, used only as necessary.

NB. For the special case of an address line broken up by unfolding, see Addresses above.

It is quite common that the last page of a letter is divided in three, with two strips of content separated by the address panel in the centre. In such cases our preferred practice is to move the address down rather than the bottom strip up, since the address is the last to be written, and ▼ in the middle of the page is easier for relating image and transcription at a glance.

Although not an objective of this project, the code linking <anchor> and <ref> would permit a hyper-diplomatic version of transcription in which everything was left in situ.

3.6.7. Footnotes: <note>[@resp]

An editorial footnote is inserted where something is to be explained to the reader. The display script moves it to the foot of the text and numbers it sequentially, leaving a lemma number in situ, raised above the line. On the project website, hyperlinks allow the reader to jump from lemma to footnote [actually an endnote] and back; in MDC, the footnotes appear just below the text corresponding to one image, and the numbering restarts with each new page. Footnotes are coded as <note> with the sole attribute @resp, to distinguish them from Annotations for the XSLT script. The value of @resp is "#XYZ", where XYZ is the note-writer's standard identifier, aka the editor's initials; the # means that the xml:id has been declared within the document.

There is hardly any TEI/XML restriction on the position of <note>. The lemma for a footnote usually follows the last character of the span of text referred to. A footnote referring to the whole document, e.g. explaining an editorial change of sequence, will normally go straight after <pb n="1"/>.

A footnote referring to a particular element usually goes just before that element's closing tag. This avoids the footnote lemma jumping to the next line if the element is part of a centred or right-aligned span. If a footnote concerns an annotation, however, that would cause the raised footnote lemma number to be displayed in green italics, like the rest of the annotation, so the footnote's opening <note> is best placed after the annotation's closing </note> — so long as there is no problem keeping the lemma on the same line; if there is, too bad, and italicising the number is the lesser of two evils.

• A footnote lemma follows a span of text including any closing punctuation, i.e. according to British rather than American publishing practice.
• There is no space before a footnote lemma, except to separate lemmas where there is more than one footnote at a particular point in the transcription.
• A footnote ends with a period.
• References to book titles in a footnote go in italics, not quotation marks.
• Quoted material goes in single curly quotation marks, but when discussing the meaning of a word, distinguish the word itself (in italics) from the meaning(s) or sense(s) attached to it (in single curly quotation marks).
• References to the OED should be laid out in one of the ways illustrated below:
  […] Tredrille<note resp="#TO">‘A card-game played by three persons, usually with thirty cards’ (<hi rend="italic">OED</hi> s.v. <hi rend="italic">tredrille</hi> n. Accessed 02-07-2020).</note> […] pugh<note resp="#TO">This reading is uncertain, but <hi rend="italic">pugh</hi> is a rare 18th-century spelling of the interjection <hi rend="italic">pooh</hi> ‘[e]xpressing impatience, contempt, disdain, etc.’ (<hi rend="italic">OED</hi> s.v., B.1. Accessed 19-06-2020).</note>

Please include a footnote when the image of a document only contains covers, in order to clarify the absence of a transcription.

NB. What this document calls ‘footnotes’ are presented as such in MDC transcriptions, but they are displayed as endnotes on the project website.

3.8.1. Style

<hi> requires @rend with usually 1 but can be up to 3 values from the list "underlined | superscript | subscript | italic | strikethrough | double-underlined | triple-underlined | dotted-underlined | wavy-underlined | larger | all-caps | over-inked-pen | retraced". The last four are ignored in display, and all the variants of "underlined" are displayed as single underline.

NB. Superscript and subscript are reasonably grouped here among style attributes, since they display text in a smaller font as well as slightly raising or lowering it.

<emph> requires @rend with between 1 and 3 values from "underlined | italic | double-underlined | triple-underlined | dotted-underlined | wavy-underlined | retraced | larger | all-caps", with similar display limitations.

Wherever appropriate, underlining should be done with <emph> rather than <hi>. This needs thorough revision, since certain values of @rend, and indeed the element <emph> itself, had not been available in the Image to Text project.

<l> (verse line) optionally allows @rend with 1 of the following values: "underlined | italic".

3.8.2. Rightwards movement

On many content elements, @rend is concerned with left-right positioning, with possible values "indent | center | align-right". On our project website, these labels are somewhat misleading. Value "indent" inserts some hard spaces to indent the first line of the element only. Values "center" and "align-right" shift the left margin inwards for the duration of the element, rather than adjusting each line of text to be genuinely centered or aligned to the right margin, as in a word processor. Good practice generally is to use @rend in this way only on an element that starts at the left margin. If used on an element with text to its left, it will invoke an undesired line-break before the element. The values "center" and "align-right" for @rend always introduce a blank line in the display after the element, which is also often undesirable. For these reasons, unless it is particularly convenient to have @rend in another element, we are moving increasingly towards using @rend in <lb>, where it applies to that line only and does not introduce blank lines before or after it.

The default layout without @rend is left alignment.

NB. It is inadvisable to insert rightwards-moving @rend in more than one element on the same line. For cases where there needs to be more than the usual inter-word spacing between items on the same line, use the <space> element instead (on which see Extra spacing between items above). Leave one normal space to the left of <space>.

3.8.3. Cancellation of midline linebreak

The last use of @rend is to handle a peculiarity of the interaction between TEI and HTML. In normal use, a paragraph <p> element in XML or HTML is treated as such in on-screen HTML, starting on a line of its own and finishing on a line of its own. However, there are situations where TEI demands an opening <p> or a closing </p> at a point which must be midline if we are to preserve the lineation of the ms. The principal cases are these:

• A postscript begins with ‘P.S.’ or similar and may then continue on the same line with the content of the postscript. TEI requires that the line start with <postscript>, followed by ‘P.S.’ wrapped in <label>, followed by the content of the postscript wrapped in <p>. The tag <p> is thus inevitably midline. We place rend="inline" in <p> to instruct the display script not to jump to a new line:
  <lb/><postscript><label>P.S.</label> <p rend="inline">I am in daily <lb/>expectation of a letter from […];
• A div break requires a paragraph break as well, and the new paragraph may not be the very first item after the new <div>. In this example, <dateline> is not allowed in <p>:
  <lb/>we have had a good deal of rain this afternoon -- </p>         </div>         <div type="continuation"> <dateline><date when="1789-07-15">15<hi rend="superscript">th</hi></hi></date>/</dateline> <p rend="inline">This morning I <lb/>got up at 5 O'Clock on purpose to write to you [...]
• To eke out the paper, a writer starts a new paragraph not on a new line but after an extra space on the same line:
  <lb/>[…] I continue to amuse myself very much, with Music. <lb/>I have been once at the Opera, and was very <lb/>well entertained.</p> <space unit="chars" quantity="5"/><p rend="inline"><persName>Lady Stormont</persName> is come <lb/>to <placeName>Town</placeName>. <rs>Her Daughter</rs> is much improved […]
• A closer always requires the preceding paragraph to end just before it, and the changeover point may well be midline:
  <lb rend="indent"/>God bleſs and preserve You -- </p> <closer rend="inline"><seg type="closeSalute">my dear <lb/>dear <persName>Mary</persName> [...]

NB. @rend is not permitted in <fw>, nor anywhere in the header, specifically in <p>.

3.9.1. <salute> vs. <seg>

References to the addressee and the author are tagged with <salute> or <seg> (and <persName> as well, if a proper name is used), depending on the role of the string in the letter:

• <salute> is a structural element containing the greeting at start or end of the letter.
• <seg> is used to tag mentions of the addressee in <p>, as well as of the author.

Final punctuation goes inside the <salute> or <seg> element. NB. This does not apply to <persName> and <rs>, neither of which includes final punctuation, only internal punctuation that is part of the reference, e.g. in an abbreviation.

If an explicit name for the addressee is given, both <salute> and <seg> take <persName>:

Pronouns such as you are not tagged either as <rs> or <seg>.

3.9.2. Referring to persons in <opener>

This will always be a reference to the addressee. If the name is given, use <persName> inside <salute>, if not, <rs> in <salute>:

3.9.3. Referring to persons in <p>

3.9.4. Referring to persons in <closer>