Documenting the Libretto

1. The Libretto’s Genetic Text Edition

1.1 Preliminary Remarks

Die digitale textgenetische Edition des Librettos basiert inhaltlich auf der gedruckten Textbuch-Ausgabe.¹ The content of the libretto’s digital genetic text edition is based on the printed libretto edition.1 Unlike the historico-critical printed edition where paramount is preparing a critically-revised work text and documenting its variants in order not only to lay the foundation for further scholarly work, but also to provide an authoritative text for music praxis, the focus of this digital text edition has been considerably expanded to emphasize genetic text processes. To be elucidated are alteration processes within a single manuscript (“text layers”), as well as also text-formation processes in the overlapping individual manuscripts extant (“text stages”) culminating in initially authoritative changes that can be observed during the course of the printed libretto’s reception.
For the concept of the genetic text edition within Freischütz Digital, this specifically means that the edition and the point of reference for documenting how the rest of the sources “deviate” is no longer based on one main source, but that all text sources (quasi equivalent) used for the edition are juxtaposed, transcribed and imaged digitally, and each is processed as necessary. The primary interest here, besides the exact description and chronological classification of the materials, is in exploring and mapping a source’s different text stages using the digital medium, together with linking the individual sources in order to classify these in the overlapping genetic text process or to show existing connections among the text levels of the various sources. A complete visualization of the genesis process requires technical measures, however, for which sufficient resources were lacking in the present project, so that only exemplary implementation was possible. But the way the texts are displayed is in place as the basis for other forms of visualizing the documented connections.

1.2 Auswahl der Textquellen

Friedrich Kind’s Freischütz libretto has been handed down in a variety of manuscript and printed texts. The digital edition takes into account all authentic librettos, that is, authorized by Kind and Weber. Compiling these sources illustrates the fact that the textual foundation of Weber's composition significantly changed over a longer period of time from the first idea (in 1817) to the premiere on 18 June 1821, and that Carl Maria von Weber was himself decisively involved in this process. Furthermore, the various printed editions made between 1822 and 1843 by the librettist on his own initiative reveal further enlightening text versions and variants. Weber, however, accepted these in part by sending the printed libretti to the theaters, thus sanctioning them and thereby attaining for them performance-practice relevance. The premiere’s actual text, on the other hand, cannot be completely reconstructed, for it is extant only in the form of the first print of the arias and songs.²

Overview (Primary Sources)

File Name	Title	Date of Writing Down	Sort of Source	Sigla	xml:id
Kind’s Manuskript	Die Jägersbraut	May 1817	Autograph of the librettist	L-tx2 (D–B)	L-tx2
Weber’s Personal Exemplar	Der Freyschütze	June 1817	Textbook Copy with autograph additions	KA-tx4(D–B)	KA-tx4
Copy for Berlin	Die Braut des Jägers / Der Freyschütz	August 1819	Textbook Copy	K-tx6 (D–B)	K-tx6
Copy for Gotha	Die Braut des Jägers	October 1819?	Textbook Copy	K-tx7 (D–GOl)	K-tx7
Copy for Vienna	[Der Freischütz]	August 1821	Textbook Copy with autograph additions	KA-tx15(D–Wn)	KA-tx15
Copy for Hamburg	Der Freyschütz	October 1821	Textbook Copy with autograph additions	KA-tx21(D–Hs)	KA-tx21
Pre-Print of the non-composed duet Eremit/Agathe	Die geweiheten Rosen	Autumn 1819	Pre-Print	VD1	VD1
Pre-Print of the Folksong No. 14	Volkslied aus der Oper: Der Freischütz	September 1820	Pre-Print	VD2	VD2
First Print of Arias and Vocal Numbers 1821	Arien und Gesänge der romantischen Oper: Der Freischütz	June 1821	First Print	ED-tx	ED-tx
1. Issue of the Textbook	Der Freischütz	End of 1821/beginning of 1822	Printed Textbook	D-tx1	D-tx1
2. Issue of the Textbook	Der Freischütz	Spring 1822	Printed Textbook	D-tx2	D-tx2
3. Issue of the Textbook	Der Freischütz	Summer 1823	Printed Textbook	D-tx3	D-tx3
Issue within Theaterschriften	Der Freischütz	Summer 1827	posthumous Print of the Textbook	D+-tx2	Dp-tx2
Last Issue (Ausgabe letzter Hand)	Der Freischütz	Spring 1843	posthumous Print of the Textbook	D+-tx3	Dp-tx3

In addition to the primary sources classified as authentic or authorized, the present edition is supplemented by a number of libretto sources of interest to the reception history, such as, for example, the libretto containing the material for the Detmold performances since 1825 (K-tx29), and a printed libretto of 1827 published under Friedrich Kind’s name (D+-tx1), together with two other manuscript librettos for the 1821 first performance in Vienna (K-tx15a and K-tx15b; neither of these is available digitalized, but only encoded in TEI).

1.3 Encoding and Visualizing the Sources

All manuscripts and printed texts are faithfully transcribed diplomatically and marked according to TEI standards (Text Encoding Initiative; www.tei-c.org/). Each source is represented by a separate XML file. Captured in the individual encodings are not only material or formal aspects (e.g., source structure and typeface, writing instruments, font, superscripts and subscripts, underlinings, typographic alignment, etc.), but the content attributes defining the kind of respective text section (dialogue text, verse lines, staging directions, names of persons, etc.) are also considered. Also reflected in the encodings is the different status of the sources (working manuscript, fair copy, copy, etc.). Each source is adequately marked in depth according to relevant objectives, i.e., specific structures and layers of writing or texts will also be captured using relevant TEI elements (for example, the intervention by known and unknown hands as substitutions).³ Whereas the extant printed librettos (first print of the songs for the premiere (UA) and the Friedrich Kind editions) reproduce, respectively, only a text that can be exactly dated, the manuscript sources combine preponderately different text levels, attributed in some instances to several authors (writers) or stemming from widely disparate chronological sections.

The diplomatically-faithful source encoding is transferred using the Edirom user interface to a reading version enabling not only exploring the text content, but also tracing step by step the editorial interpretation of the text genesis. Edirom visualizes in the process the reconstructed or defined individual text layers within a manuscript, by, for example, displaying and hiding the various stages of text formation or synoptically displaying them juxtaposed with other sources to reveal in turn different or dependent relationships among the sources. (As noted, this is a first visualization realized with limited technical outlay to demonstrate the options; based on the encodings more comfortable presentations could be implemented at any time.)

From the Edirom source window, various views of the texts are possible:

1. Metadata (including source description)
2. Continuous text (flow text)
3. Page-based facsimile
4. Page-based text – facsimile juxtaposition
5. XML data view
6. Synopsis with other sources (with variance display)

---

2. Concept of Editing the Genetic Text Edition

The genetic text edition focuses on presenting the genetic text processes within the individual source (referred to here as “text layers”) and in direct source comparison (as “variances” here or as a succession of “text stages”). The actual text development is to be understood as a transparent process with disparate, in part only fragmentarily documented stages.
In transmitting the text its development is made transparent by conceiving it in layers. Via the menu all separable text layers by identified writers can be selected individually as well as displayed in the synoptic juxtaposition. Then again source comparison can be done in two different ways: the synopsis of two (selectable) sources and the variance display within the text transcriptions. The variance display enables the overview of all (meaningful) variances of the sources edited.
The conception of this edition is due to crucial ideas of the groundbreaking work for the genetic text edition,⁴, but the terms “text layer,” “variance” and “text stage” were used here in a form specific to the project.

2.1 Displaying the Single-Source Text Layer

The conception of displaying the text layer within Freischütz Digital was influenced, inter alia, by developments from software development, for the difference between two files is frequently established there by means of a diff tool (difference tool) – a display form transferred in a modified form to the present subject matter. Presenting the text layers this way is relevant only for the manuscript sources. The prints always reproduce only one text version imaged accordingly in the edition.
There are two ways to display text layers in the manuscripts:

Reproducing the source in separate text layers

The text layers can be selected individually via Text layers in the menu. The number of text layers displayed here usually depends on the number of clearly differentiated writers, thus corresponding essentially to a simplifying overview of the writing layers characterizing the manuscript, whereas an internal differentiation (for instance, of a copyist’s individual writing acts) was not undertaken. Writers who could not be unambiguously identified were classified under the heading “anonymous.” Identifying the text layers is done subject to the entries’ respectively implicit chronological order, in which the text-layer concept is interpreted as additive, i.e., a new text layer is made up of the previously valid text components, now newly established (by erasure, replacement, or addition), thus relevant to the overall valid text at the point in time selected. The individually selectable text layers display the text version selected as “clear-text,” i.e., as diplomatically-transcribed running text without indicating interventions vis-à-vis the preceding text version.

Reproducing the source with all interventions within the “fictive” final layer

The reproduction of the text with all the interventions elucidated can be selected in the demo via “Genesis of the Text with all Intermediate Results” in the menu. Mapped here are only all the textual, content-differentiated layers where interventions are to be found; i.e., a new layer is not displayed when nothing has changed at a spot of interest as compared with that chronologically preceding it. Each interpretable correction process receives
The synoptic display of text layers is done using various (either predefined or user-selectable) colors to color-code the relevant text line. Given this way is additional information about the correction process:

• 1. Information about the respective writer (writer siglum)
• 2. Classifying the kind of process (+ for addition, - for deletion, -/+ for replacement)
• 3. Verbal explanations of the correction process (form and placement of the intervention)

For use of the various editorial genetic-text options, see Tools (Demos).

2.2 Presenting the Genetic Text (Source Variance, Text-Stage Chronology)

In addition to displaying the text layers of the individual sources, clarifying correlations and/or distinctions among the primary sources and the libretto becomes a major focus of the genetic-text edition, illustrating in so doing the process of text formation and transformation. Since there are no readings vis-à-vis a prepared edited text, but all sources are equivalent as evidence of the continuous alteration process, the variances observed must be used to make this process tangible.

Variance Concept

Only meaningful variances, i.e., substantial deviations among the sources were considered in the present edition. Differences in punctuation and orthography were ignored since each individual source is completely encoded as well as also available.
In addition, the capturing of “variances” was restricted to those interventions going back to or authorized by one of the work’s two authors (Friedrich Kind as librettist and Carl Maria von Weber, crucially involved as composer in the text formation). Anonymous corrections that could not be attributed, together with corrections done in conjunction with local performance versions (e.g., within copies for Vienna and Hamburg) are conversely ignored in the variance display. These differences can, however, be traced in the text-layer display of the relevant sources in individual steps.

Encoding Variances

The differences (variances) are not captured in encoding the individual sources, but have been centrally collected as a stand-off markup within an external, so-called Core File, broken down in turn by scenes. Using a tool, the Core-Builder especially developed for this by Raffaele Viglianti (Maryland Institute for Technology in the Humanities), the relations among individual sections, verses, or words and/or word groups requisite for the edition could be produced.
All variants are listed in the core file in the form of individual apparatus entries and thus appear similar to a traditional list of variants or readings. Nevertheless, there are fundamental differences in implementing this method: First of all, the sections, lines, or words pertaining to the respective source encodings are defined and identified by individual identification numbers (ID’s). From the core file, these (automatically assigned) XML identifiers are then referenced.
The apparatus entries then refer to these ID’s, i.e., they contain the source references in the form of relevant <rdg/> ("reading") elements with the attribute @wit (“witness”), initially containing the ID of the XML file given the source. Several <rdg/> ("readings") can be combined into groups <rdggrp/> ("reading-group"). The individual <rdg/> elements then refer to the ID of the relevant single spot of the single-source file by the element <ptr/> ("pointer") and the attribute @target. If a variance has no correspondence to a single source or to several sources, the <ptr/> is missing in the <rdg/> element.
The apparatus entries are additionally categorized according to the variance type (omission, addition, substitution, conversion, spelling). Combinations may also occur.

Example for a machine entry in the core file (with two reading groups)::

<app type="substitution">
    <rdggrp n="1">
        <rdg wit="#freidi-librettoSource_KA-tx4_xml_id_d1e9293">
            <ptr target="#d1e9409"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_L-tx2_xml_id_d1e10504">
            <ptr target="#d1e10654"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_K-tx6_xml_id_d1e9223">
            <ptr target="#d1e9337"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_K-tx7_xml_id_d1e8345">
            <ptr target="#d1e8452"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_KA-tx15_xml_id_d1e10326">
            <ptr target="#d1e10459"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_KA-tx21_xml_id_d1e10144">
            <ptr target="#d1e10285"/>
        </rdg>
    </rdggrp>
    <rdggrp n="2">
        <rdg wit="#freidi-librettoSource_D-tx1_xml_id_d1e8196">
            <ptr target="#d1e8297"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_D-tx2_xml_id_d1e8185">
            <ptr target="#d1e8291"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_D-tx3_xml_id_d1e8130">
            <ptr target="#d1e8236"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_Dp-tx2_xml_id_d1e8085">
            <ptr target="#d1e8190"/>
        </rdg>
        <rdg wit="#freidi-librettoSource_Dp-tx3_xml_id_d1e8488">
            <ptr target="#d1e8588"/>
        </rdg>
    </rdggrp>
</app>

2.3 Variance-Display Options

Variance display relating to single-source text transcriptions

This variance display can be done from any source window (but only in the “text” view) and shows the corresponding source differences of all other primary sources defined by the core files in comparison to the libretto
In more complex manuscript sources with several text layers, it should be noted that selecting the variance in the core files is based on only the respective, last chronological text layer (thus the text version attained in this layer) to be viewed as authentic and/or authorized by both of the work’s authors, Kind and Weber [cf. here under variance concept].

This way of defining the text layer as the display basis pertains to the following sources (in this version the text is displayed in the text transcription):

• L-tx2, Kind autograph (second text layer, including all Kind’s corrections)
• KA-tx4, personal exemplar (the text version including Weber’s interventions and entries by copyist 2, though without the later Jähns entries)
• K-tx6, Berlin copy (text version including corrections by the Berlin copyist)
• KA-tx15, Vienna copy (text version including Weber’s corrections, though without other corrections)
• KA-tx21, Hamburg copy (text version including Weber’s corrections, though without other corrections)

The variance markers can also be displayed, hidden, and individually controlled via the menu. Selecting and opening a window gives an overview of the meaningful variants captured for the relevant text item in the core file. Displayed behind the source sigla listed in the table are the relevant source texts in the corresponding text transcription; the identical readings are grouped here and listed together. Taken into account for sorting the group were also the development of the text passage in each individual source.
If necessary, navigation from the text window to the other sources not currently open is possible in order to view the passage in its context. The machine entries listed in the core file can also be displayed in the XML view. If no single-passage reference is possible because text passages do not exist in individual sources, a reference is made to the source’s entire source file.

Displaying variance with the synopsis of two optionally selected sources

The variances captured by the core file can also be displayed in the juxtaposition of two optionally selected primary sources in the text transcription via “View” and “Two-Text Synopsis” in the menu. Appearing in the left column of the text is the source opened in the text transcription, downloaded in the right column can be the text transcription of any other text. Again, the restriction to only one (authentic or authorized) text layer applies. After making the selection, the differences between the two selected texts can be switched on or off via the menu. These are displayed and can be controlled in the text by different icons (for the various icons, see Variance Icons and Comments).

2.4 Additional Comments and Further Links to WeGA

Editorial comments referring only to one source are inserted as <note type=“commentary“/> in the relevant individual source’s encoding. Multiple source annotations are collected (similar to variances in a higher-level XML file), sorted by scene, and categorized by content. There are general comments, comments on the text constitution, comments defining words, and comments referring to Weber’s diaries and writings [see encoding guidelines].
All <note/> elements refer to their sources within the <note/> attribute @target. The linking is based on specifying the source sigla in connection with the xml:id of the respective placement in the individual sources.
In the Edirom interface, the comments to be displayed or hidden are indicated by various icons [see Variance Icon and Comments]. When clicking on an icon the comment will be visible using a tooltip.
References to contents in the Carl-Maria-von-Weber-Gesamtausgabe [Complete Edition] (http://www.weber-gesamtausgabe.de) (e.g., for the diary) are also included in the separate commentary file, in which each reference is provided with a <ref/> element, giving in the attribute @target the relevant item’s URL address (a diary page, as an example = http://weber-gesamtausgabe.de/A060183).

2.5 Variance and Comments Icons

• Icon code-fork rotated = variance display (content of the core files by scene)
• Icon comments-o = source-overlapping comments (pertaining to multiple sources)
• Icon comment-o = single-source comments
• Book = reference to compositional dates from Weber’s diary
• Note = text variance in Weber’s score autograph
• Asterisk = footnote text (found only in the prints)

---

3. Encoding Guidelines

3.1 Preliminary Comments

The editorial guidelines of the Freischütz Digital project’s libretto group are based on the editorial guidelines of the Carl-Maria-von-Weber-Gesamtausgabe – Digitale Edition (see http://www.weber-gesamtausgabe.de/de/Editionsrichtlinien) s), together with the guidelines of the Text Encoding Initiative (TEI) in version P5 (see TEI-Guidelines). The encoding guidelines relate to both the libretto’s primary sources as well as also to the edition’s references sources. Issues specific to reference sources are explained accordingly. It should be noted that the most comprehensive information is always found in the encoded text itself, i.e., the rendering of the encoded texts in “human-readable” form (i.e., the HTML presentation) generally contains the information only in a reduced or “conditioned” form, i.e., in filtered and reordered form).

3.2 Principles of Transcribing and Presenting Texts

Structuring

All XML documents consist of two fields: the metadata (information about the file and its contents) and the actual text corpus. The text corpus (= <text/>) is in turn subdivided into a sector of the libretto’s upstream information such as endpapers, title pages, and casting list (= <front/>)/>), followed by the “pure” libretto text (= <body/>). Text markup is TEI-specific according to formal and content-related aspects. Using the TEI-specific labeling, attempts are made to capture the structure of the individual sources according to their appearance. Targeted is taking over the model diplomatically (including the original orthography and punctuation). The coarse structure of the texts (<body/>) is captured using classifications by act and scene (= <div type=“act“/> and <div type=“scene“/>). The fine structure is dealt with by marking the dramatis personae (<speaker/>)/>), the paragraphs (<p/>) as dialogue-text indicators, the verse lines (<l/> or <lg/>), and stage directions (<stage/>).The placing of text sections is indicated by attribute (@rend = “left,” “center,” and “right,” @rend = “indent,” etc.) [note: the prefixed @ is a sign for attribute]. The indentations of dialogue or verse lines, consistently carried out especially in the libretto’s manuscript primary sources, are unified in the text transcription, or partly ignored. Page breaks are indicated by the element <pb/>with information about the page number and folio numbering (count attribute @n = “1v,” etc.; pagination or foliation as per the source). The original typographic alignment is captured by the element <lb/> at the start of the line (the mostly prevalent justified type of text-alignment is, however, ignored in the text transcription).

Metadata

Structuring the metadata takes place in the <teiHeader/> subdivided into three sections: <fileDesc/>, <profileDesc/> and <revisionDesc/>. The <fileDesc/> is a description of the contents of this data set (title, accessibility, source description – <sourceDesc/> etc.). Following in the <profileDesc/> is information about the text’s (historical) “profile” (genesis, date, type of text, language, etc.). The <revisionDesc/> contains information on the revision of the current data set (documentation of the changes).

Overall framework information (file declaration)

DThe first line of an XML document contains information about the XML version and the character encoding used, the second line refers to the schema underlying the file. Finally, the third line identifies the text as a TEI document (indicating the corresponding namespace) and assigns a unique identification number xml:id) for this document.

Reproducing the manuscript texts

Shortcut lines (also gemination lines / duplication lines) are tacitly written out in doubled letters (in the case of shortcut lines added later the second of the consonants is defined as <add>). Hyphens at line breaks are not adopted (only within the <lb>-element is the existing hyphen defined by <lb @break=“no“ @rend=“-“> definiert). Text sections in superscript or subscript, underlining, etc., are explicitly indicated in the encoding (<hi rend=“superscript“/>, <hi rend=“subscript“/>, <hi rend=“underline“ n=“1“/> etc.). Upper- and lower-case in the model are faithfully reproduced (even in the use of upper-case letters within single words). Writing or printing errors in the source are generally adopted and marked by the element <sic/>. If corrected by the editor, this is done in combination with the <sic/> element (marking the original spelling using the <sic/> element, correcting with <corr/>; including both in a <choice/> element). Text sections in Latin within a text written in normal typeface are marked using the element <hi/> (highlighted) and the attribute @rend=“latintype” (representation in the displayed text is usually a cursive as opposed to non-cursive typeface). Bracketed characters “/:” and “:/” were adopted unchanged. Original punctuation marks are adopted; in exceptional cases, a sign supplied by Hg [Ed] may be inserted in the form of <supplied>,</supplied>. Editorial interventions were made sparingly (mittels <supplied/>).

Lines or line units within paragraph (<p/>), shared in the sources by two or more persons (usually indicating simultaneous speech), are marked by <seg part="I"> (initial), <seg part="M"> (middle) und <seg part="F"> (final).

Transcribing the printed texts

Reproducing different fonts in the case of printed texts was carried out by marking them as <hi rend=“antiqua“>, <hi rend=“bold“>, <hi rend=“spaced_out> etc.. Original footnotes/annotations are reproduced through the combined elements <ref/> (referencing string) and <footnote/>.

Dealing with text interventions (tracing text genesis)

Deletions of words or paragraphs are marked by the element <del/> (deleted), the deletion form is indicated by the attribute @rend (rendition) (@rend=“strikethrough”) or by the attribute @hand to specify more closely the respectively identified writer. Later inserts are noted with the element <add/> (addition); the insertion location is indicated by the attribute @rend: “inline,” “margin,” “above” (the line), “below” (the line), “top” or “bottom,” and also specifies the writer. Larger sections deleted are marked with the element <delSpan/>, overlapping structural additions using <addSpan/> and then likewise provided with the necessary attributes. The end of the individual deletions and additions is marked in this case by the element <anchor/>. Authorial corrections or those by a copyist are represented as a substitution process within the element <subst/> (usually of two parts: the deletion process (through deletions, overwriting, or suchlike) and the addition of new text (through insertion, overwriting, or suchlike), thus by a combination of <del/> and <add/>. Overwriting is likewise marked as a substitution process, occurs by deleting through “overwriting” and by adding “inline.” The illegible is reproduced with editorial additions (example: <damage agent="water"><unclear>...</unclear></damage>)or the completely unreadable with <gap/>.

Additionally in the reference sources:

If the various interventions (deletions and additions) are made in various writing colors and the respective writer is unknown, the various writing colors are indicated in the encoding as @type=“ink” for interventions in ink, @type=“pencil” for interventions in pencil, and @type=“redpencil” for interventions in red crayon.

Marking text contents

Marking the role names occurs directly via <name/> and <rs type=“name“/> or using the more general reference element <rs/> ("referencing string"), more closely specifying in the attribute the type of target: <rs type="person"/>.

Comment forms

Editorial comments are captured by the element <note/>, or more closely specified by the attribute @type: Possible <note/> attributes are: @commentary: pertains to explanations of facts otherwise not directly comprehensible, @textConst: text constitution; refers to remarks on formal peculiarities of the source, as well as @definition: includes pure terminology or conceptual explanations.

Keywords in the Topic Maps

In order to ensure the interrelationship of motifs, terms, and names, a marking of the respective keywords in the text is necessary. The keywords are encoded in the XML encodings with for names and with <name/> for names and with <rs/> for terms, each provided with an @key attribute and with an identification referring in turn to an external Topic Map, separate from all texts.

Marking text quotations in the reference texts

Marking verbal quotations from external texts or within the text edition is done by <cit/>, and further subdivides into <quote/> for the actual quotation, and into <bibl/> with the bibliographic information on the text’s origin. When the quoted text lines or passages are adopted unaltered from the quotation source, the <cit/>is marked with @type=“true”, and if the quotation has been altered, by @type=“modified.”

---

4. Selected Research Results within the Project Framework

4.1 Interdependent Relationship of the Berlin and Gotha Copies as well as the Kind Printed Editions on a Lost Libretto Manuscript

Because of the broad agreement of the Berlin copy K-tx6 and the Gotha copy K-tx7, especially the identical differences from KA-tx4, the Schreiter libretto edition (p. 190) already posited that it was not the Weber personal exemplar, but another, no longer extant source that served as model for these libretto copies. This could be the copy for Friedrich Kind [cf. Source Overview, K-tx5] that, according to his diary of 21 June 1817, Weber had had prepared. That the two copies K-tx6 and K-tx7 are directly related to the lost manuscript [K-tx5] is suggested by numerous correspondences with Kind’s later printed editions (D-tx1–3 as well as D+-tx2 and 3, cf. Table 1). For it can be assumed that these, again, are presumably based on the libretto remaining in Kind’s possession, which continued covertly to change up to the text version reflected in the printed editions [cf. here also the text development from D-tx1 to D+-tx3 under the text development in the Kind printed editions].

Examples	KA-tx4 und all other Sources	K-tx6, K-tx7 as well as Printed Editions by Kind
No. 1 Introduction 2nd Vers in Kilian’s Song	„Stern und Straus hab’ ich vor’m Leibe“	„Stern und Straus trag’ ich vor’m Leibe“
Dialogue I/2	„meinen Jägerpurschen“	„meinen Purschen“
	„in jenen Zeiten“	„in alten Zeiten“
Dialogue I/5	„Das will ich dir lernen.“	„Das will ich Dich lehren.“
No. 10 Finale	„Caspar. Nicht? So bleib […] zu thun, Hexenmeister?“	missing (later added in K-tx6)
Dialogue III/3	„Brautthränen und Frühregen währen nicht lange, sagt das Sprichwort.“	„Brautthränen und Frühregen, sagt das Sprichwort [in D+-tx3 Sprüchwort], währen nicht lange.“
No. 16 Finale	„mit einem finstern Blick“	„mit finsterm Blick“

Since some of the differences from Weber’s personal exemplar found in K-tx6 and K-tx7 as well as the Kind printed editions also show up again in the first edition of the vocal texts (ED-tx) for the premiere (UA), we can definitely assume that ED-tx is based, inter alia, on the Berlin libretto copy (K-tx6) (cf. table 2). This means that passing these minor differing details along from the beginnings of the work’s genesis in 1817 until the UA in 1821 can only be explained by linking Kind’s lost manuscript [K-tx5] to the first Berlin libretto (K-tx6) dependent on it.

Example	KA-tx4 and all other sources	K-tx6, K-tx7 as well as Printed Editions by Kind
No. 2 Terzetto with Chorus	Max. Nimmer trüg’ ich den Verlust	missing
No. 9 Terzetto	„der, der Gott versucht“	„der, wer Gott versucht“ (also in L-tx2)
No. 10 Finale	„Ja – furchtbar gähnt“	„Ha – furchtbar gähnt“
No. 15 Hunter’s Chorus	„Was glich wohl auf Erden“	„Was gleicht wohl auf Erden“
No. 16 Finale	„die eigne Braut!“	„die Braut!“
	„war immer treu und gut!“	„war immer brav und gut!“
	„Heil unserm Fürst!“	„Heil unserm Herrn!“

4.2 Similarities and Differences between Weber’s Personal Exemplar and Friedrich Kind’s Printed Editions

Examples	KA-tx4	D-tx1 to D+-tx3
		Hermit’s scenes again included
Dialogue I/2 and 4
No. 2 Terzetto with Chorus	Kuno, /.faßt Max bei der Hand./ Mein Sohn! nur Muth! Wer Gott vertraut, baut gut!	taken over
Dialogue I/5 and 7	Caspar Das würde sich das Wachspüppchen das mich um deinetwillen verwarf, schwerlich einbilden. für sich. Es soll gerochen werden!!	taken over
No. 6 Duetto	Agathe. Wer bezwingt des Busens Schlagen? Wer der Liebe süßen Schmerz? Stets um dich Geliebter zagen Muß dieß ahnungsvolle Herz.	Agathe. Wer bezwingt des Busens Schlagen? Wer der Liebe süßen Schmerz? Annchen. Die bezwingen Lust und Scherz! Agathe. Stets um den Geliebten zagen Muß dieß ahnungsvolle Herz!
No. 10 Finale – Chorus of Ghosts		slightly modified
Melodrama		taken over
	„reicht ihm die Jagdflasche, die Max verweigert.“	„wirft ihm die Jagdflasche zu, die Max weglegt.“ (correction by Weber revised)
No. 12 Cavatina	Cancelling of verse 3	taken over
No. 14 Volkslied		Weber’s corrections included
Dialogue III/5	Ein Todtenkranz!	Eine Todtenkrone!
No. 15 Hunter’s Chorus	Verse 2	taken over
No. 16 Finale	10 lines cancelled in the text of the hermit	Only 4 lines cancelled

4.3 Classifying and Evaluating the Detmold Libretto Copy

In the case of the 2007 libretto edition, a source to be more closely considered here was not yet extant. There is a manuscript libretto copy, now in the Lippische Landesbibliothek in Detmold (siglum: Mus-n 245, Landesbibliothek Detmold stamp on fol. 2r). It was hitherto surmised that this libretto is related to the extant score copy (K-pt15) likewise in Detmold, the score provided for the first performance (EA) in Bremen on 26 September 1822 by the theatre entrepreneur August Pichler who took over direction of the Detmold court theatre in 1825.
The libretto, however, seems to have a more complex provenance.
The first puzzle is about its origin. It was written down by two persons, one of them was Friedrich Langer, listed as owner on the wrapper, from whom stems the beginning of the manuscript up until fol. 12r; the other one, Hugo Langer, continued the copy (fols. 12v to 73r) and dated it in a note on fol. 73r, “Geyer the 1st of August 1823 Scripsit Hugo Langer.”
The libretto copy’s content is based on Friedrich Kind’s 2nd libretto edition from Göschen/Leipzig in 1822, proven by the existence of the opening hermit scenes, as well as by the faithful adoption of this edition’s footnotes. Yet nevertheless the libretto must still have been intended for a musical performance, as suggested in turn by the prominent entry “Opening […] music.” before the beginning of the 3rd scene (actually the opera’s 1st scene) as well as the numerous deletions and the remarks in pencil/blue pencil and red crayon referring to music numbers.
Whether and for which of the early performances of “Freischütz” the libretto was consulted, currently cannot be reconstructed due to the few entries.
Further evidence of provenance is provided by the personnel entries in the casting list, together with another existent theatre stamp “DIRECTION DES II. THEATERS IN DRESDEN” (fols. 2r and 72v), though these raise more questions than provide answers. The 2nd Dresden theater is the summer theater, built in the Great Garden in 1856 by Joseph Ferdinand Müller, called Nesmüller; its performances were of a more popular character, but nevertheless a competition to the Royal Theater not to be underestimated. Most of the folk plays done there were by Nesmüller himself, inter alia, “Die Soldatenfamilie,” “Die Frau Tante,” or “Die Zillerthaler.” In 1881, theater had to close down due to financial difficulties (see Das alte Dresden, Bilder und Dokumente aus zwei Jahrhunderten, eds. Erich Haenel and Eugen Kalkschmidt, reprint of the 1934 edition, F./M., 1977). Only two of the cast entries have been clearly attributed, namely, “Boxberg” pertaining to the Ottokar role, and “Pechtel,” pertaining to Max. The former is Eugen von Boxberg, the actor and regisseur whose appearance, according to the stage almanac [Bühnen-Almanach]], can first be traced to Greiz in 1862, and then to several other (stage) locations, inter alia, in Hof, Bayreuth, Annaberg, Pirna, Großenhain, and Grimma, where he played paternal and official roles. Theodor Pechtel’s career, in turn, verifiably started in 1863 in Döbeln and led him to Milwaukee and Chicago via, inter alia, Hischberg, Flensburg, and Hamburg. Pechtel played at the 2nd Dresden Theater for only the one year, 1869. Since there is no proof at all that Boxberg played at the Dresden Theater, a direct connection between the two persons seems scarcely to be possible, and it remains extremely puzzling why both names are on a casting list. Still (so much is certain), Boxberg and Nesmüller, the Dresden theater director, did cross paths, though a decade too late to be connected with Pechtel. According to the 1878/79 Bühnen-Almanach, Nesmüller’s ensemble made guest appearances at the theater company where Boxberg was in charge at this time, in Grossenhain, Pirna, und Oschatz. In 1878 a third name on the list, “Rothe,” assigned to both Cuno and Samiel, can also be verified for the 2nd Dresden stage. Anyhow, it can be assumed that in this way the libretto might possibly have fallen into Nesmüller’s hands and hence reached the holdings of the 2nd Dresden Theater. From there, how it came (again?) to Detmold remains to date unexplained. Also revealing are the performance-practice entries in the libretto. These interventions implemented (either in different phases or from different hands) involve numerous cuts to the dialogue text and scenes, as well as, above all, major changes in the music numbers (e.g., cuts in no. 1, deleting the Bohemian Waltz and Max’s aria, no. 3; cuts in no. 2 /retrospectively cancelled; deletion, 1st strophe of the Trinklied [Drinking Song]/ 3rd strophe, censorship variant added later; deletion of Max’s aria in no. 10, finale, 2nd act, as well as other cuts in the subsequent dialogue and deletion of the dreadful song; deletion of the 3rd and 4th strophes of the bridal wreath, no. 14; deletion of the 2nd strophe, hunters’ chorus, no. 15; several deletions in the finale, 3rd act).
To what extent the Detmold (Dresden) libretto is related to the Bremen score copy (presently Detmold = K-pt15 and partly duplicate extant material still possibly going back to two holdings), must still be investigated in more detail; the annotation on the alteration of the act structure – Wolfsschlucht [Wolf’s Glen scene] as the start of Act 3 – corresponds with a portion of the Detmold material; but the text difference in the 2nd strophe in Caspar’s Trinklied, no. 4, shows significant difference in the two materials (see for this, the next section).

4.4 Developing the Text and the Relationships of the Text Underlay within the Music Sources (Example, Trinklied, 2nd Strophe)

There are hardly any significant differences in the vocal texts very consistently passed down through the extant score copies, unlike that of the text in the libretto sources.
A distinctive and somewhat mysterious difference is found, however, in no. 4, Caspar’s Trinklied, in the 2nd strophe.
In three of the surviving scores (K-pt13, Weimar; K-pt15, Bremen; and K-pt20, Neustrelitz) there is, unlike in the other music sources, a difference of two lines in another text. Instead of “Kartenspiel und Würfellust | Und ein Kind mit runder Brust” it reads “Würfellust und Kartenspiel | Und ein Kind das uns gefiel.” Within the libretto sources, this text change shows up only in Weber’s personal exemplar, implemented there by copyist 2 in a subsequent correction made at an uncertain timepoint, though a chronological connection between his entry and the altered text in the scores can be conjectured. The three score copies with the new version of the text can be precisely dated from Weber’s diary comments (Weimar and Bremen invoices, 22 and 27 March 1822, respectively; Neustrelitz, payment on 16 June 1822). Also informative is the fact that the copyist’s text part at this spot was already left blank in the Stuttgart copy (KA-pt26), paid for on 27 December 1821, and that Weber himself added the text of the 2nd strophe here, though – and this is what’s interesting – in the original version. That could mean that in December 1821 Weber planned to change the text at this spot (perhaps even externally inspired), but then decided again for the old version in the subsequent score copies (extant through the three exemplars mentioned), then, on the contrary, favored the new text variant. From this, it can certainly be gathered that this text was sung in the performances at Weimar (EA, 4 May 1822), Bremen (EA, 26 September 1822), and Neustrelitz (EA, 12 August 1822, and perhaps even in still other performances from this time, of which unfortunately no evidence remains extant), but not beyond that, since it is to be found in neither the piano reductions nor in the opera’s printed scores. Only two more text sources can still be established: the 2nd issue of the libretto from Reclam Leipzig, ca. 1900, and the Wedekind edition, Berlin, 1925. Unfortunately, no further knowledge could be gained from this edition about the materials on which it was based.

4.5 Evaluating the Stage Directions at the End of Hamburg Libretto Copy (KA-tx21)

The Hamburg libretto copy was the basis for the EA in Hamburg on 5 February 1822. Unlike the two early librettos for Berlin and Gotha, the Hamburg libretto copy represents the text version after the UA (as of October 1821), including all the interventions made in the work up to then, such as for example, the hermit passages that were cut in the finale of the 3rd act; see also the description of the sources.
Of performance-history interest are the page-long set of stage and scene directions located at the end of the libretto copy (fols. 58v–70v). These performance materials are now presented for the first time within the Freischütz Digital“ edition in a diplomatic text transcription with juxtaposed facsimile. Evaluating its contents provided the following results:
These are set and stage directions in triplicate by various writers, each arranged by act and related to the libretto page numbers, supplemented by names of the respective acting personnel of the Hamburg theater, together with special directions for the Wolfsschlucht [Wolf’s Glen] scene.
The first and third sets of directions by act (fols. 58v–60r and 67r–70r, respectively) stem in fact from different unknown, but connected writers. Added for the first time as of fol. 67r in the repeat copy based on the first set was the personnel information included in the text mainly on a regular basis, or supplemented by other essential data beyond that in the first set; see especially the Wolfsschlucht scene and 3rd act. Moreover, other additions were still made in pencil (possibly in another hand).
The second set by act (fols. 63r–66r) as well as the directions for the Wolfsschlucht (fols. 61r–62r) stem from another hand, most likely from Friedrich Ludwig Schmidt, the Hamburg stage manager.

The appendix sets of directions have not been dated in their entirety, making their exact chronological classification difficult at first glance. By reconciling the personnel entries within the scene directions with the extant Hamburg address books (see the Staats- und Universitätsbibliothek Hamburg http://www.sub.uni-hamburg.de or URL: http://agora.sub.uni-hamburg.de/subhh-adress/digbib/start)), it is possible to speculate, though, that the notes belong to various performance periods and can also be narrowed down (the annual address books refer to the season beginning the previous year):

First set [version] of scene directions:

The actual set of directions is difficult to date, for it contains no personnel information. The later entries of persons in thicker ink by the same writer presumably pertain, however, only to the performances as of 1827/1828, since the stage technician or theater master Hü(h)nerjäger first shows up in the address books as of 1828.

Wolfsschlucht scene and second set of scene directions:

The notes most likely prepared by Friedrich Ludwig Schmidt pertain relatively unambiguously to the EA in Hamburg, 5 February 1822, since the Mssrs. Brämer, Petersen, and Schäfer were only part of the personnel in 1821/1822 (Brämer also still in 1822/1823); thereafter they are missing in the address books. These leaves were presumably later bound inadvertently between the two other sets belonging together chronologically.

Third set of scene directions:

This set of directions is presumably also classified like the first in the 1827/1828 season (Herr Kleinschmidt who is deleted in the list, is mentioned in the address book only until 1828; bridesmaid, Dlle. Regler, Jr., only in the 1827/1828 season; Dlle. Gröger is there as of the 1828/1829 season). But the corrections stem in part from ca. 1836 (corrections for the bridesmaid: Dlle. Pleß is there only as of autumn 1834, Dlle. Teubner, only in the autumn of 1836).

What is the significance of these sets of directions? The three different sets of directions by act are point-by-point information on personnel management, props, plot sequences, etc. The libretto contains specific instructions, scene-by-scene, for the theater personnel and also gives a partial impression of how lavish the Freischütz production’s personnel outlay was, also, inter alia, the exact listing of the necessary supernumeraries for the performance, which could be reduced by, for example, converting four of the six “mountain musicians” in the first act to “hunting cavaliers” in the 3rd act (fol. 60v). The arrangements for the Wolfsschlucht “Wie die Besorgungen in der Wolfsschlucht folgen.” (fols. 61r–62r) are exemplary and revealing evidence of how the central Wolfsschlucht scene devised by Weber and Kind (scene II/4–6) was thus implemented and equipped for a contemporary stage to be as “spine-chilling” as possible for the theater audience. The assigments for the specific personnel are listed in detail: who has to carry out which actions with which props, e.g., that Herr Brämer was responsible for ringing the bell, and the movements of the owl controlled by Herr Hollmann, Jr., were to be accompanied by the actions of Herr Woltereck, the Casper singer (fol. 61r). “When Caspar blows on the fire, the owl waggles its wings on the branch; it also turns its head to Caspar several times, when he turns around backwards,” states the applicable scene directions (fol. 64r). From corrections, it can be seen that the original “cave” was transformed into a “machine” that opened and disappeared, or that the “Wagen [wagons]” were corrected to “Irrlichtern [will o’ the wisps].” Furthermore, the stage activities were precisely defined right down to the individual echoes during bullet-casting, such as, for example, “Donner und Blitz | Sturm [Thunder and Lightning | Storm]” “Tannen brechen [breaking fir trees],” “Peitschenknall, Geratter [crack of a whip, clatter],” “Knallerbse [firecrackers], ” together with inserting the most disparate animals (lizard, snake, bear, etc.). That the effects here also increased from echo “one” to “seven” is evident, as growing out of the initial “dull tempest […] is the strongest storm with all thunder and lightning and impact,” detailed in the scene directions (fol. 65r): “At the same time, rain, persistent lightning, all thunder machines like thunderstorms battling each other – impact. Two pieces of rock fall down in the back. Will o’ the wisps dance in between storms bells in the distance. The storm is steadily raging.” Weber also gives this direction in his “Remarks on the Scenic Instructions for Freyschütz.” (originating for the performance in Munich on 15 April 1822); see publication in the Freischütz-Edition, 2007, pp. 244ff., and http://weber-gesamtausgabe.de/A031689). Since Friedrich Ludwig Schmidt himself verifiably attended the Berlin performance [see Denkwürdigkeiten des Schauspielers, Schauspieldichters und Schauspieldirektors [Memorabilia of the Actor, Playwright, and Theater Director] Friedrich Ludwig Schmidt (1772–1841)), compiled from posthumous drafts and edited by Hermann Uhde, Stuttgart, 1878, vol. 2, p. 169], he was able to get an accurate impression of the scenic design (especially that of the Wolfsschlucht scene) implemented in line with Weber’s ideas. The “Freischütz” was just as successful in Hamburg as in Berlin and elsewhere, as Schmidt summed up further (ibid., p. 193): “For the Epiphany [in 1823] we gave the ‘Freischütz’ for the first time in the new year; since February 5, 1822, it has by now been performed twenty-eight times within eleven months, thus on the average of every twelve days, and to ever-increasing applause. The first year of its appearance in Hamburg, the net income for this opera amounted to 26,281 Marks 3 Schillings; the average intake for each performance was therefore 938 Marks; from the sale of 1,766 libretti (at six Schillings each) we had still gotten an extra 661 Marks 14 Schillings.”

4.6 Text Development in the Kind Printed Editions and its Significance in Relation to the Text Handed Down in the Music Sources (Including the Interpretation of the Most Important Scores and Librettos up to the Start of the 20th Century)

The printed editions of the librettos published by Friedrich Kind are revealing with respect to the genesis of the text. These editions, beginning with the first at the turn of 1821/1822 to the last in 1843, show in a clear manner the librettist’s further work on the text; he had suffered under his position as subordinate to the successful Freischütz composer, and intended to distance himself from former, joint decisions, such as, for example, by deleting the opera’s originally planned opening scenes that Weber intentionally wanted dropped. By purchasing the libretto in March 1817, Weber had secured the property rights to the book for five years, so that Kind himself could not publish the text until 1822. In contrast to his own manuscript and the text in Weber’s personal exemplar, Friedrich Kind made numerous changes (apparently arbitrarily) within these various editions. In addition, the editions also vary substantially among themselves. It is interesting to note, however, that Weber obviously tolerated the changes made in the Kind edition, for he sent the cheaper printed books to the theaters starting with the turn of the year 1821/1822 instead of the copies produced until then by copyists. Hence, within the reception of the Freischütz libretto, or in conveying the text in the opera’s contemporary and subsequent performances, the early Kind editions published up to Weber’s death are of considerable significance; albeit a detailed investigation of the performance-practice relevance of these librettos is still pending.⁹ Among the most serious of the alterations in the Kind editions in comparison with Weber’s personal exemplar are, besides the restoration of the opening hermit scenes, also the reduction of the hermit passage by four lines in the finale of the 3rd act (in contrast to Weber’s authorized ten-line cut), and the omission of the Ännchen aria, no. 13, in the two late editions (Theaterschriften [Theater Writings] of 1827 and the final definitive edition of 1843). It cannot be a coincidence that Kind brought about the deletion of no. 13 only after Weber’s death. Although the aria was composed especially for the UA, it had become established as an integral part of the work.

	1. Aufl. 1821/22	2. Aufl. 1822	3. Aufl. 1823	Theaterschriften 1827	Ausgabe letzter Hand
introductory scenes of the hermit	+	+	+ (with footnote in the text)	+ (with commentary in the Epilogue)	+ (with commentary in the Epilogue)
Cuno’s Romanza	-	-	-	+ (in the Epilogue)	-
Ännchen’s Aria	+	+	+	-	-
Cut of the hermit’s passage by 4 lines (Finale Act III)	+ (with marking the following six lines)	+ (with marking the following six lines)	+ (without marking)	+ (without marking)	+ (without marking)

---

Footnotes

1. Friedrich Kind/Carl Maria von Weber, Der Freischütz. Critical Libretto Edition by Solveig Schreiter, Munich, 2007. ↩

2. Johannes Kepper, Solveig Schreiter, Joachim Veit, Freischütz analog oder digital – Editionsformen im Spannungsfeld von Wissenschaft und Praxis, in: editio 28 (2014), pp. 127–150; Solveig Schreiter: Zur Quellensituation des Freischütz-Librettos, in: Perspektiven der Edition musikdramatischer Texte. Conference Report Bayreuth University, 22.–24. November 2012 (in progress)↩

3. Encoding Guidelines, Text Edition.↩

4. To be mentioned at this point are merely two fundamental publications: Almuth Grésillon, Éléments de critique génétique, Paris , 1994, and the anthology, Textgenetische Edition, eds. Hans Zeller and Gunter Martens, supplement to editio, vol. 10 (1998). ↩

5. Herrmann Abert, Carl Maria von Weber und sein Freischütz, quoted from: Der Freischütz. Texte, Materialien, Kommentare, p. 170↩

6. Susanne Balhar: Das Schicksaldrama im 19. Jahrhundert. Variationen eines romantischen Modells, Frankfurt, 2004, pp. 48f.↩

7. Cf. here also Kathrin Wulfhorst: Intertextuelle Bezüge in den Freischütz-Texten von Johann Apel und Friedrich Kind, in: Weberiana 19 (2009), pp. 101–124↩

8. Cf. here also ibid., pp. 103-104, and Schreiter, Der Freischütz, pp. 103ff.↩

9. Solveig Schreiter and Joachim Veit, Zu Quellensituation und Referenzen des Freischütz-Librettos, Perspektiven der Edition musikdramatischer Texte (wie Anm. 2, in progress)↩