Declaring witnesses in TEI critical editions

🚧 🚧 Documentation under construction — This page and its subpages are currently being revised. Contributions are welcome: feel free to correct, expand, and improve them.

Guide 2 of 6 — Declaring witnesses in TEI critical editions

Previous: Understanding TEI documents for critical editions  ·  Collection overview  ·  Glossary  ·  Next: Encoding a basic critical apparatus in TEI

A critical edition depends on textual evidence.

Before an apparatus can state that one witness reads soul while another reads mind, the witnesses themselves must be identified.

This guide extends the TEI document created in Guide 1.

The document will move from:

TEI
├── teiHeader
│   └── file description
└── text
    └── body
        └── paragraph

to:

TEI
├── teiHeader
│   └── fileDesc
│       └── sourceDesc
│           └── listWit
│               ├── witness A
│               ├── witness B
│               └── witness C
└── text
    └── body
        └── paragraph

The progression followed in this guide is:

textual source
      ↓
witness
      ↓
witness declaration
      ↓
XML identifier
      ↓
editorial siglum
      ↓
reference from later apparatus data

At the end of the guide, the TEI file will contain a small but coherent witness list that can later be cited from lemmas and readings.

1. Where this guide fits in the collection

The six-guide workflow is:

TEI document
      ↓
[WITNESS DECLARATIONS]        ← Guide 2
      ↓
basic apparatus entries
      ↓
complex textual variation
      ↓
Lua processing
      ↓
ConTeXt typesetting

Guide 1 established:

• the TEI root element;
• the TEI namespace;
• the TEI header;
• the textual body;
• a paragraph with an XML identifier.

This guide adds:

• the concept of a textual witness;
• witness declarations;
• a witness list;
• stable XML identifiers;
• reader-facing sigla;
• descriptions of manuscripts and printed editions;
• the distinction between declarations and references;
• the basic structure that apparatus entries will later use.

This guide does not yet explain:

• <app>;
• <lem>;
• <rdg>;
• the detailed encoding of witness agreement;
• complex witness families;
• Lua-based validation;
• final apparatus typography.

Those subjects belong to later guides.

2. Understanding textual witnesses

2.1. What is a witness?

A witness is a source that preserves or transmits a form of the text.

Depending on the work being edited, a witness may be:

• a manuscript;
• a printed edition;
• a papyrus;
• an inscription;
• a typescript;
• an authorial draft;
• a historical transcription;
• a quotation in another work;
• a translation;
• another identifiable textual source.

Suppose that the sentence used in Guide 1 survives in three sources:

A  The mind seeks unity.
B  The soul seeks unity.
C  The mind desires unity.

The three lines are not merely three copies of the same file.

They represent three textual witnesses:

textual tradition
├── witness A
├── witness B
└── witness C

Each witness has its own identity, history, and textual content.

A critical edition compares those witnesses and records their agreements and differences.

2.2. Witness, source, and edition

The words witness, source, and edition are related but should not be treated as automatic synonyms.

Term	Principal meaning in this collection
Source	Material from which textual or editorial information is derived
Witness	An identifiable source that preserves or transmits a form of the text
Edited text	The text established and presented by the editor
Critical edition	The complete scholarly publication containing the edited text and its documentation

A printed book may function as a witness when its readings are included in the textual comparison.

A modern critical edition may also be used as a secondary source without being treated as one of the primary witnesses.

The status of a source therefore depends partly on the editorial project.

3. Declarations and references

3.1. Why witnesses must be declared

A printed apparatus may use compact sigla:

A B C

To a human reader, these may appear sufficient.

A structured edition, however, must know what those sigla identify.

A witness declaration can connect:

machine-readable identity
        ↓
reader-facing siglum
        ↓
witness type
        ↓
description
        ↓
later references

For example:

XML identifier:  ms-A
printed siglum:  A
type:            manuscript
description:     the principal manuscript

Without declarations, a reference such as:

wit="#ms-A"

has no defined target.

The declaration establishes that target.

Guiding principle. Declare each witness once, then refer to it by its stable XML identifier. The declaration stores identity and description; later references reuse that information without duplicating it.

3.2. The declaration-and-reference model

The witness system contains two complementary operations.

3.2.1. Declaration

A declaration creates and describes an identifiable witness:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

3.2.2. Reference

A later element can refer to that witness:

wit="#ms-A"

The relation is:

declaration
xml:id="ms-A"
        ↑
        │
reference
#ms-A

The declaration stores the identity.

The reference points back to it.

This principle will later allow apparatus entries to remain compact and consistent.

4. Building the witness list

4.1. The witness list

TEI provides the element <listWit> for a list of witnesses.

A small witness list may look like this:

<listWit>
  <witness xml:id="ms-A" n="A">
    The principal manuscript.
  </witness>
  <witness xml:id="ms-B" n="B">
    A second manuscript containing several alternative readings.
  </witness>
  <witness xml:id="ed-C" n="C">
    An early printed edition.
  </witness>
</listWit>

Its structure is:

listWit
├── witness
│   ├── xml:id: ms-A
│   ├── n: A
│   └── description
├── witness
│   ├── xml:id: ms-B
│   ├── n: B
│   └── description
└── witness
    ├── xml:id: ed-C
    ├── n: C
    └── description

The witness list acts as a small registry.

Each witness is declared once and can be referenced many times.

4.2. Where the witness list belongs

In the document used by these guides, the witness list is placed inside the source description of the TEI header:

teiHeader
└── fileDesc
    ├── titleStmt
    ├── publicationStmt
    └── sourceDesc
        └── listWit
            ├── witness
            ├── witness
            └── witness

The relevant source fragment is:

<sourceDesc>
  <listWit>
    <witness xml:id="ms-A" n="A">
      The principal manuscript.
    </witness>
    <witness xml:id="ms-B" n="B">
      A second manuscript containing several alternative readings.
    </witness>
    <witness xml:id="ed-C" n="C">
      An early printed edition.
    </witness>
  </listWit>
</sourceDesc>

This placement reflects the role of the witness list:

TEI header
    ↓
documentation of the encoded resource
    ↓
description of its sources
    ↓
declaration of textual witnesses

4.3. The <listWit> element

The TEI element <listWit> groups witness declarations.

In the examples used here, it contains several <witness> elements:

<listWit>
  <witness>...</witness>
  <witness>...</witness>
  <witness>...</witness>
</listWit>

The list has two principal functions:

• it gives the witness declarations a common location;
• it allows processing systems to retrieve the declared witnesses together.

Conceptually:

listWit
└── witness registry
    ├── A
    ├── B
    └── C

A larger project may use several lists or more elaborate source descriptions. The present guide keeps the structure deliberately small.

4.4. The <witness> element

Each <witness> element declares one witness.

For example:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

This declaration contains three distinct parts:

Part	Example	Function
Element	<witness>	Declares a textual witness
XML identifier	xml:id="ms-A"	Provides a stable machine-readable identity
Editorial label	n="A"	Supplies a short value that can be used as a siglum
Content	The principal manuscript.	Describes the witness

The same model is repeated for each witness.

5. Identifiers, sigla, and descriptions

5.1. XML identifier and siglum are not the same thing

This distinction is fundamental.

Consider:

<witness xml:id="ms-paris-123" n="P">
  Paris, Bibliothèque nationale de France, manuscript 123.
</witness>

The declaration contains:

XML identifier:  ms-paris-123
siglum:          P
description:     Paris, Bibliothèque nationale de France,
                 manuscript 123

The identifier and siglum serve different audiences.

XML identifier	Siglum
Used for machine-readable linking	Used for compact scholarly display
Must follow XML identifier rules	Follows the editorial conventions of the edition
Should remain stable in the source	May vary between publications
Need not be elegant	Should be concise and readable
Example: ms-paris-123	Example: P

The distinction can be visualised as:

<witness xml:id="ms-paris-123" n="P">
          └──────────────┘    └─┘
           machine identity   display siglum
</witness>

A project may choose identical values:

xml:id="A"
n="A"

but that does not make their functions identical.

Do not confuse identity with display. The value of xml:id is a linking target inside the XML document. The siglum is the compact label presented to readers. They may look alike, but they serve different purposes.

5.2. Choosing stable XML identifiers

For these guides, witness identifiers should:

• be unique within the document;
• begin with a letter or underscore;
• contain no spaces;
• remain stable when possible;
• identify the witness rather than its current visual form;
• avoid depending on page layout or apparatus order.

Useful identifiers include:

ms-A
ms-B
ed-C
papyrus-46
ms-paris-123
print-basel-1781
translation-latin-01

Less useful identifiers include:

first-witness
red-siglum
bottom-note-source
page-7-manuscript
temporary-B

The second group describes a temporary role or visual appearance rather than a stable identity.

5.3. Choosing witness sigla

A siglum is a compact editorial label.

Common sigla include:

A
B
C
P
M
P46
Vat.
α
β

A useful siglum should normally be:

• short;
• distinguishable from other sigla;
• stable within the edition;
• documented in the witness list;
• suitable for repeated use in the apparatus.

The siglum belongs to the presentation of the scholarly data.

For example:

machine reference:  #ms-paris-123
printed siglum:     P

A later ConTeXt process can resolve the reference and retrieve the siglum:

#ms-paris-123
      ↓
find xml:id="ms-paris-123"
      ↓
retrieve n="P"
      ↓
print P

5.4. Describing different witness types

The witness content should give the reader enough information to understand what the witness represents.

A simple manuscript declaration might be:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

A second manuscript:

<witness xml:id="ms-B" n="B">
  A later manuscript containing several alternative readings.
</witness>

A printed edition:

<witness xml:id="ed-C" n="C">
  An early printed edition.
</witness>

A fuller project may record substantially more information, but the basic distinction is already visible:

Identifier	Siglum	Type represented by the description
ms-A	A	Principal manuscript
ms-B	B	Later manuscript
ed-C	C	Printed edition

The prefixes used here are project conventions:

ms-     manuscript
ed-     printed edition

TEI does not require these particular prefixes.

They simply make the identifiers easier to understand.

5.5. A first witness list

The minimal witness list for this guide is:

<listWit>
  <witness xml:id="ms-A" n="A">
    The principal manuscript.
  </witness>
  <witness xml:id="ms-B" n="B">
    A later manuscript containing several alternative readings.
  </witness>
  <witness xml:id="ed-C" n="C">
    An early printed edition.
  </witness>
</listWit>

The corresponding editorial inventory is:

Siglum	XML identifier	Description
A	ms-A	The principal manuscript
B	ms-B	A later manuscript containing several alternative readings
C	ed-C	An early printed edition

The same information appears in three forms:

editorial inventory
        ↓
TEI witness declarations
        ↓
later apparatus references

6. Extending the TEI document

6.1. Extending the TEI file from Guide 1

In Guide 1, the source description contained only a short paragraph:

<sourceDesc>
  <p>Created for the ConTeXt Garden TEI guides.</p>
</sourceDesc>

Replace that paragraph with the witness list:

<sourceDesc>
  <listWit>
    <witness xml:id="ms-A" n="A">
      The principal manuscript.
    </witness>
    <witness xml:id="ms-B" n="B">
      A later manuscript containing several alternative readings.
    </witness>
    <witness xml:id="ed-C" n="C">
      An early printed edition.
    </witness>
  </listWit>
</sourceDesc>

The document now has this structure:

TEI
├── teiHeader
│   └── fileDesc
│       ├── titleStmt
│       ├── publicationStmt
│       └── sourceDesc
│           └── listWit
│               ├── witness [xml:id="ms-A", n="A"]
│               ├── witness [xml:id="ms-B", n="B"]
│               └── witness [xml:id="ed-C", n="C"]
└── text
    └── body
        └── p [xml:id="p1"]

6.2. Complete TEI MWE

Save the following document as:

tei-guide-02.xml

<?xml version="1.0" encoding="UTF-8"?>

<TEI xmlns="http://www.tei-c.org/ns/1.0">
  <teiHeader>
    <fileDesc>
      <titleStmt>
        <title>A small critical-edition example</title>
      </titleStmt>

      <publicationStmt>
        <p>Unpublished teaching example.</p>
      </publicationStmt>

      <sourceDesc>
        <listWit>
          <witness xml:id="ms-A" n="A">
            The principal manuscript.
          </witness>

          <witness xml:id="ms-B" n="B">
            A later manuscript containing several alternative readings.
          </witness>

          <witness xml:id="ed-C" n="C">
            An early printed edition.
          </witness>
        </listWit>
      </sourceDesc>
    </fileDesc>
  </teiHeader>

  <text>
    <body>
      <p xml:id="p1">The mind seeks unity.</p>
    </body>
  </text>
</TEI>

This document contains:

• one TEI root;
• one TEI header;
• one file description;
• one source description;
• one witness list;
• three witness declarations;
• one textual body;
• one identifiable paragraph.

It does not yet contain an apparatus entry.

That addition belongs to Guide 3.

6.3. Reading the witness declarations line by line

Consider the first declaration:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

Source fragment	Function
<witness>	Declares one textual witness
xml:id="ms-A"	Assigns a unique XML identifier
n="A"	Supplies a concise editorial label
The principal manuscript.	Describes the witness
</witness>	Closes the declaration

The declaration can also be represented as:

witness
├── machine identity: ms-A
├── display siglum: A
└── description: The principal manuscript.

7. Reusing witness declarations

7.1. How apparatus entries will refer to witnesses

Guide 3 will introduce elements such as:

<lem wit="#ms-A">mind</lem>
<rdg wit="#ms-B">soul</rdg>

These lines are not yet part of the Guide 2 MWE.

They are shown here only to explain the purpose of the declarations.

The references resolve as follows:

wit="#ms-A"
      ↓
find xml:id="ms-A"
      ↓
retrieve n="A"
      ↓
display A

and:

wit="#ms-B"
      ↓
find xml:id="ms-B"
      ↓
retrieve n="B"
      ↓
display B

The complete relation will be:

witness declarations
├── ms-A → A
├── ms-B → B
└── ed-C → C

apparatus references
├── #ms-A
├── #ms-B
└── #ed-C

7.2. One declaration, many references

A witness should normally be declared once.

It may then be referenced from many apparatus entries:

declaration
xml:id="ms-B"
      ↑
      ├── reading in apparatus entry 1
      ├── reading in apparatus entry 2
      ├── reading in apparatus entry 3
      └── reading in apparatus entry 4

This avoids repeating the full description every time the witness is cited.

Instead of storing:

soul — A later manuscript containing several alternative readings

inside every apparatus entry, the source can store a reference:

wit="#ms-B"

The declaration supplies the reusable information.

7.3. Referencing several witnesses

When several witnesses support the same reading, TEI references can be written as a space-separated list:

wit="#ms-B #ed-C"

Conceptually:

one reading
├── supported by ms-B
└── supported by ed-C

The processing system must therefore distinguish:

attribute value:
    "#ms-B #ed-C"

individual references:
    #ms-B
    #ed-C

resolved sigla:
    B
    C

Guide 3 will show this structure inside an actual reading.

7.4. Declaration order and display order

The order of declarations in the witness list does not necessarily have to determine the final printed order.

For example, the source may declare:

A
B
C

while an edition may require:

A C B

or group witnesses by type:

manuscripts: A B
printed editions: C

Three orders should therefore be distinguished:

Order	Meaning
Declaration order	Order in which witnesses occur in the TEI source
Editorial order	Order established by the edition’s scholarly conventions
Printed order	Order used in a particular output

A small project may use the same order for all three.

A larger project may define an explicit sorting rule with Lua.

That processing belongs principally to Guide 5.

Order is a separate editorial decision. Declaration order records how witnesses occur in the source file; editorial and printed order may later be established independently.

8. Extending the witness model

8.1. Witness type and witness identity

A witness type classifies the witness.

A witness identity distinguishes one particular witness from another.

For example:

type: manuscript
identity: ms-A

type: manuscript
identity: ms-B

type: printed edition
identity: ed-C

Two witnesses may share a type while remaining distinct:

manuscript
├── ms-A
└── ms-B

The identity must therefore not be replaced by the type.

This would be insufficient:

xml:id="manuscript"

when two different manuscripts exist.

Each witness needs its own unique identifier.

8.2. Simple and fuller declarations

The declarations in the MWE are deliberately small:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

A real project may eventually record more information, such as:

• repository;
• shelfmark;
• date;
• material;
• extent;
• provenance;
• scribal hands;
• bibliographical description;
• relation to other witnesses.

The conceptual progression is:

simple declaration
      ↓
stable identifier and siglum
      ↓
fuller source description
      ↓
project-specific manuscript metadata

The present guide establishes the first two levels.

It does not attempt to replace a full TEI manuscript description.

8.3. Witness groups and families

Witnesses may sometimes be associated with a textual family, recension, or other editorial group.

For example:

family α
├── A
└── B

printed tradition
└── C

A group may be useful for:

• analysis;
• filtering;
• ordering;
• apparatus abbreviation;
• reporting shared readings.

However, a group should not erase the identities of its members.

The following information remains distinct:

individual witness identity
        +
group membership

For example:

A belongs to family α
B belongs to family α
C belongs to the printed tradition

The exact TEI representation of complex witness groups depends on the project’s encoding model and selected schema.

This guide therefore keeps the MWE at the level of individual declarations.

Grouping and project-specific classification can be developed after the basic reference system is stable.

9. Validation and common errors

9.1. Consistency rules for witness declarations

A small witness registry should satisfy the following rules:

Every witness has one unique xml:id.
Every witness has a documented siglum or display label.
Every later witness reference points to a declaration.
No two declarations use the same xml:id.
The same witness is not declared repeatedly under different identifiers
without an explicit editorial reason.

These rules can be summarised as:

Rule	Reason
One unique identifier per witness	Prevents ambiguous references
One declaration per witness	Avoids duplicated metadata
Stable identifiers	Keeps references valid during revision
Documented sigla	Makes printed output understandable
All references resolve	Prevents undeclared witnesses from appearing in the apparatus

9.2. Common declaration errors

9.2.1. Duplicate identifiers

Incorrect:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>
<witness xml:id="ms-A" n="B">
  A later manuscript.
</witness>

Both witnesses use:

xml:id="ms-A"

The identifier is no longer unique.

Correct:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>
<witness xml:id="ms-B" n="B">
  A later manuscript.
</witness>

9.2.2. Reusing one identifier for a type

Incorrect:

<witness xml:id="manuscript" n="A">...</witness>
<witness xml:id="manuscript" n="B">...</witness>

The word manuscript describes a type, not the identity of each object.

Correct:

<witness xml:id="ms-A" n="A">...</witness>
<witness xml:id="ms-B" n="B">...</witness>

9.2.3. Confusing the reference with the identifier

Declaration:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

Correct later reference:

#ms-A

The number sign belongs to the reference, not to the identifier declaration.

Incorrect declaration:

<witness xml:id="#ms-A" n="A">
  The principal manuscript.
</witness>

Correct relation:

xml:id="ms-A"
      ↑
      │
reference="#ms-A"

9.2.4. Using the siglum as an unexplained string

A later apparatus could contain:

B

as an ordinary character.

But this does not create a structured connection with the witness declaration.

A reference such as:

wit="#ms-B"

preserves the link:

apparatus reading
        ↓
witness declaration
        ↓
siglum and metadata

9.3. Well-formedness and witness declarations

The witness list must still follow the XML rules established in Guide 1.

Correct:

<listWit>
  <witness xml:id="ms-A" n="A">
    The principal manuscript.
  </witness>
</listWit>

Incorrect, because the element is not closed:

<listWit>
  <witness xml:id="ms-A" n="A">
    The principal manuscript.
</listWit>

Incorrect, because the attribute value is not quoted:

<witness xml:id=ms-A n=A>
  The principal manuscript.
</witness>

Incorrect, because the nesting crosses:

<sourceDesc>
  <listWit>
</sourceDesc>
  </listWit>

The proper structure is:

sourceDesc
└── listWit
    └── witness

The closing order must be reversed:

open sourceDesc
open listWit
open witness
close witness
close listWit
close sourceDesc

9.4. Structural validity and editorial correctness

A witness declaration may be structurally valid but editorially wrong.

For example:

<witness xml:id="ms-A" n="A">
  A fifteenth-century manuscript.
</witness>

may be valid XML and valid TEI.

But if the manuscript actually dates from the thirteenth century, the declaration is editorially inaccurate.

Three levels must therefore remain distinct:

A valid file can still be wrong. XML well-formedness and TEI validity cannot confirm that a shelfmark, date, siglum, or witness description is historically accurate. Editorial verification remains necessary.

Level	Question
Well-formed XML	Are the XML tags and attributes syntactically correct?
Valid TEI	Is the witness declaration permitted by the selected TEI model?
Editorial correctness	Does the declaration describe the witness accurately?

10. Inspecting the witness list with ConTeXt

The complete XML source used by this test was created in section 6.2 and saved as tei-guide-02.xml.

The final typesetting of witness sigla and apparatus entries belongs to Guide 6.

A small ConTeXt test can nevertheless verify that the witness declarations contained in tei-guide-02.xml are accessible.

Practical note. Save tei-guide-02.xml and tei-guide-02.tex in the same directory. The ConTeXt file loads the XML source directly with \xmlprocessfile.

Save the ConTeXt file as:

tei-guide-02.tex

Use:

\xmlregisterns
  {tei}
  {http://www.tei-c.org/ns/1.0}

\startxmlsetups xml:tei:document
  \xmlsetsetup
    {#1}
    {tei:TEI|tei:teiHeader|tei:fileDesc|tei:sourceDesc|tei:listWit}
    {xml:tei:flush}

  \xmlsetsetup
    {#1}
    {tei:titleStmt|tei:publicationStmt|tei:text}
    {xml:tei:ignore}

  \xmlsetsetup
    {#1}
    {tei:witness}
    {xml:tei:witness}
\stopxmlsetups

\xmlregistersetup{xml:tei:document}

\startxmlsetups xml:tei:flush
  \xmlflush{#1}
\stopxmlsetups

\startxmlsetups xml:tei:ignore
  % Content ignored in this witness-list test.
\stopxmlsetups

\startxmlsetups xml:tei:witness
  \noindent
  \bold{\xmlatt{#1}{n}}
  \quad
  \xmlflush{#1}
  \par
\stopxmlsetups

\starttext

\subject{Declared witnesses}

\xmlprocessfile
  {tei}
  {tei-guide-02.xml}
  {}

\stoptext

The intended output is (The PNG output isn't shown here because there are only these three lines):

Declared witnesses

A  The principal manuscript.

B  A later manuscript containing several alternative readings.

C  An early printed edition.

The processing path is:

tei-guide-02.xml
        ↓
ConTeXt loads the TEI document
        ↓
the TEI namespace is associated with the prefix tei
        ↓
the witness elements are selected
        ↓
the n attribute is retrieved
        ↓
the witness description is flushed
        ↓
a readable witness list is typeset

This is still only an inspection test.

It does not yet:

• process apparatus entries;
• resolve @wit references;
• sort witnesses;
• group witness families;
• validate missing declarations;
• apply final siglum typography.

10.1. The role of the TEI namespace

The XML source declares:

xmlns="http://www.tei-c.org/ns/1.0"

The ConTeXt file registers a shorter processing prefix:

\xmlregisterns
  {tei}
  {http://www.tei-c.org/ns/1.0}

This allows selectors such as:

tei:witness

to identify TEI witness elements.

The relation is:

TEI namespace URI
http://www.tei-c.org/ns/1.0
        ↓
ConTeXt processing prefix
tei
        ↓
selector
tei:witness

The prefix used in the ConTeXt file is a local processing convention.

It does not change the XML source.

10.2. Retrieving the siglum

The witness declaration contains:

n="A"

The ConTeXt setup retrieves it with:

\xmlatt{#1}{n}

Conceptually:

<witness xml:id="ms-A" n="A">
                         ↑
                         │
                retrieve attribute n
                         ↓
                         A

The setup then typesets it in bold:

\bold{\xmlatt{#1}{n}}

This bold form is only a demonstration.

The final appearance of witness sigla will be defined in the typesetting guide.

10.3. Source data and display remain separate

The XML source stores:

<witness xml:id="ms-A" n="A">
  The principal manuscript.
</witness>

The ConTeXt setup controls the display:

\startxmlsetups xml:tei:witness
  \noindent
  \bold{\xmlatt{#1}{n}}
  \quad
  \xmlflush{#1}
  \par
\stopxmlsetups

The separation is:

TEI XML	ConTeXt
Declares the witness	Selects the witness
Stores its identifier	Retrieves attributes
Stores its siglum value	Determines siglum typography
Stores its description	Determines paragraph formatting

A change from bold sigla to small capitals does not require the TEI witness declarations to be rewritten.

11. A practical witness checklist

Before moving to Guide 3, verify the following.

Check	Expected result
Witness list	One <listWit> in the source description
Witness count	Three <witness> elements
Identifiers	ms-A, ms-B, and ed-C
Identifier uniqueness	Each xml:id occurs once
Sigla	A, B, and C
Descriptions	Every witness has readable descriptive content
Namespace	The TEI namespace remains declared on <TEI>
XML structure	All elements are properly nested and closed
ConTeXt inspection	The three witnesses can be displayed from the XML source

The expected witness tree is:

sourceDesc
└── listWit
    ├── witness
    │   ├── xml:id: ms-A
    │   ├── n: A
    │   └── The principal manuscript.
    ├── witness
    │   ├── xml:id: ms-B
    │   ├── n: B
    │   └── A later manuscript containing several alternative readings.
    └── witness
        ├── xml:id: ed-C
        ├── n: C
        └── An early printed edition.

12. What this guide has established

This guide has moved from unidentified textual sources to a reusable witness registry:

textual sources
        ↓
individual witnesses
        ↓
TEI witness declarations
        ↓
stable XML identifiers
        ↓
editorial sigla and descriptions
        ↓
references available to later apparatus entries

The central distinction is between a declaration and a reference. A declaration identifies and describes a witness once; a reference points back to that declaration wherever the witness supports a lemma or reading.

The TEI source now has this structure:

TEI
├── teiHeader
│   └── fileDesc
│       └── sourceDesc
│           └── listWit
│               ├── witness A
│               ├── witness B
│               └── witness C
└── text
    └── body
        └── paragraph

The next guide will use these declarations inside the first apparatus entry.

13. Next guide

The witnesses are now declared, identified, and available for reuse. Guide 3 introduces <app>, <lem>, <rdg>, and @wit, and shows how readings are connected to the witnesses that support them.

---

Guide 2 of 6 — Declaring witnesses in TEI critical editions

Previous: Understanding TEI documents for critical editions  ·  Collection overview  ·  Glossary  ·  Next: Encoding a basic critical apparatus in TEI

Related pages

• Building critical editions from TEI XML
• Understanding TEI documents for critical editions
• Glossary of terms used in the TEI XML critical edition guides
• TEI XML
• XML
• ConTeXt and Lua programming
• Processing XML with Lua