Digitized Publications of Spirituals from the Nineteenth Century

Compilers and editors of hymnals and scholars of hymnology have often lacked suitable tools for identifying the earliest sources of spirituals, or even key sources that serve as models for later arrangements. In the twenty-first century, the development of internet-based repositories of digital books has enabled the ability to search for publications of spirituals using strings of lyrics or keywords, but more importantly, these repositories allow researchers to examine the relevant sources and glean contextual information about those spirituals beyond what might exist in any list or index. Although African slaves had been present in North America since 1619, this unique musical artform was not considered a national treasure worth preserving and publishing until the onset of the Civil War, thus any study of sources of antebellum plantation spirituals really begins at the end of that era and moves forward from there. In order to understand the problem and the digital solution to tracing these songs, a brief overview of the longstanding publishing standard will be presented, followed by an overview of older research materials, then a detailed examination of three existing repositories (HathiTrust, Google, Internet Archive), and one forthcoming repository (Sounding Spirit). The publications located in these repositories have been tied together through a pair of web-based bibliographies at Hymnology Archive, covering the years 1862–1900 and 1901–1942.

Background

There has been a longstanding practice in the music industry of publishing spirituals under the simple attribution ‘Spiritual’, ‘Negro Spiritual’, ‘African American Spiritual’, or something similar. While this might seem like an obvious choice, it runs counter to how other American folk songs and tunes are often attributed. This is especially true in the realm of hymnal publishing.

For example, in the hymnal Glory to God (2013), produced for the Presbyterian Church, USA, the song ‘Let us break bread together on our knees’, No. 525, was labelled as an African American Spiritual, arranged by Melva Wilson Costen (1933–2023); while on the other hand, the American folk tune NETTLETON, No. 475, was labelled as being from Wyeth's Repository of Sacred Music, Part Second (1813), or the tune AMAZING GRACE (NEW BRITAIN), No. 649, was labelled as being from Columbian Harmony (1829). A similar scenario can be seen in the Lutheran Service Book (2006) of the Lutheran Church Missouri Synod. There we find ‘There is a balm in Gilead’, No. 749, labelled as an African American Spiritual, while WONDROUS LOVE was labelled as being from Southern Harmony (1835). In other words, anonymous songs from this beloved African American musical tradition have typically been labelled according to their genre, whereas anonymous songs from European American traditions are often labelled according to their earliest known source.¹

To be fair, folk songs and tunes are not always labelled by their sources. In the Lutheran Service Book, the famous Irish tune SLANE, No. 738, is simply ‘Irish’; KINGSFOLD, No. 444, is ‘English’; and BUNESSAN, No. 789, is ‘Gaelic, 19th cent’. Nevertheless, the traditional practice of generically labelling spirituals is worth reevaluating, and to some extent, this is being addressed in newer hymnological sources. For example, in Glory to God: A Companion (2016), edited by experienced hymnist Carl P. Daw Jr.,² the history of ‘Let us break bread together on our knees’ was given proper treatment. Daw identified the earliest source for the spiritual as an article in Journal of American Folk-Lore (1925), where the song had appeared on pages 237–8, text only.³ The author or editor of the article claimed to have gathered the materials from students of the Penn School in 1923. Daw's assessment included a thoughtful argument for possible liturgical roots, influenced by practices in the Episcopal church (including kneeling at Communion), and he even suggested the line ‘with my face to the rising sun’ could be a nod to the architectural convention of building churches with the altar at the east, under a large window.

Similarly, in the Lutheran Service Book: Companion to the Hymns (2019), whose musicological research was edited mainly by Joseph Herl, we find a detailed assessment of the roots of the spiritual ‘There is a balm in Gilead’, with the earliest source named as Folk Songs of the American Negro (1907). Both the text and music were influenced by or adapted from earlier sources, as far back as John Newton's hymn ‘How lost was my condition’ in Olney Hymns (1779).⁴

Some other recent significant attempts at tracing the histories of spirituals have been made for scholarly websites such as the Canterbury Dictionary of Hymnology, the United Methodist History of Hymns blog, and Hymnology Archive.⁵

A Brief Overview of Historical Tools

The historical lack of source attributions for spirituals likely stems from a corresponding lack of tools for editors and publishers to make those determinations. In traditional English hymnology, the practice of documenting sources in hymnal companions and other reference tools dates to the middle of the nineteenth century, to books like Joseph Belcher's Historical Sketches of Hymns, Their Writers, and Their Influence (1859), Louis Biggs’ Hymns Ancient & Modern … with Annotations, Originals, and References (1867), or John Julian's monumental A Dictionary of Hymnology (1892). Texts and tunes of hymns have been indexed in resources like Nicholas Temperley's Hymn Tune Index (in print since 1998 and online since 2000), or the Dictionary of North American Hymnology, a microfilm resource, which eventually became the foundation for the indexes at Hymnary.org. But such tools for the study of spirituals are scarce. Although some spirituals are indexed in Hymnary insofar as they appear in hymnals, Hymnary has not yet taken up the task of indexing collections of spirituals, with only one exception, namely, the National Baptist Convention's National Jubilee Melodies (1916), which contains 161 songs, most of which are traditional spirituals.⁶

One key resource, which seems to have been largely unknown or unused by hymnologists for many years, was the careful indexing of spirituals carried out by Eileen Southern and Josephine Wright in African-American Traditions in Song, Sermon, Tale, and Dance, 1600s–1920: An Annotated Bibliography, part of The Greenwood Encyclopedia of Black Music (1990).⁷ The compilers indexed every publication of spirituals known to them, ranging from 1864 to 1920, with or without music, but not including single copies published as broadsides or folios. In the internet age, the absence of a web-based version has probably diminished its influence and stifled its wider adoption, and the advancement of scholarship has revealed additional sources they had overlooked.

Another important index for tracing spirituals, via recorded performances of them, is the comprehensive discography Blues and Gospel Records, 1890–1943, 4th ed. (1997). This also is not available in a web-based resource, but some early recordings are searchable in databases like the Discography of American Historical Recordings run by the University of California Santa Barbara (UCSB), The Spirituals Database edited by Randye Jones, or the crowd-sourced site Discogs, and some recordings have been digitized via UCSB, Document Records, Internet Archive, and other companies.⁸

Digital Repositories of Published Spirituals

Three major repositories of digital books now allow for extensive primary-source research, thus potentially opening the door for better histories and accountings of spirituals. These are Google Books, HathiTrust, and Internet Archive. All three share overlapping collections of books scanned by Google, then ported to the other platforms. For the purpose of understanding the full extent of the availability of spirituals published in the nineteenth century on these digital platforms, I have assembled a pair of detailed bibliographies, which are available from the website Hymnology Archive.⁹ Each bibliographic entry is linked to these three platforms whenever a digital copy was available. The bibliography was based primarily on the work of Southern and Wright, whose book is not otherwise available digitally, and their control numbers are included in the list. Additional sources were identified via a bibliography given in William Fisher's Seventy Negro Spirituals, extensive lists in the Negro Year Book, and in Monroe N. Work's A Bibliography of the Negro in Africa and America.¹⁰ The total number of sources in the two lists on Hymnology Archive for which digital copies could be expected (sources more than 95 years old, fully in the public domain) includes 159 items, covering periodicals, books and scores.

Among the most important items in the website's linked bibliography is the first formal collection of spirituals, Slave Songs of the United States (1867), which includes an important first-hand account of the transcription process and observations on the performance practice of the singers from whom the songs were gathered. Some of that preliminary work had been described by one of the book's editors, William Frances Allen, in an article published in 1865, which is also in the website bibliography.¹¹ Two major dynasties of published spirituals were established by Fisk University starting in 1872, and Hampton University starting in 1874, both of which extended into multiple publications and editions over several decades. Some other historically Black schools are represented as well, including Tuskegee in Alabama, the Calhoun Colored School in Alabama, and the Penn School on St. Helena Island, Georgia.

Before examining the digital availability of these resources, a couple of caveats are in order. First, these lists do not include spirituals published individually as broadsides, folios or choral octavos.¹² Some of these individual pieces can be found in digital repositories, in the ones listed below and elsewhere,¹³ but a systematic approach was not attempted for this study. Second, the bibliographies at Hymnology Archive include an array of materials, all of which include texts or music scores of spirituals, which could have appeared in historical or analytical books or articles about spirituals or about antebellum slave life, in focused groups or collections of songs, or in autobiographical memoirs. The diversity of materials aids in the contextualization of where and how the songs were collected and how they were used, interpreted, and disseminated.

General Results from the Big Three Repositories

Out of the 159 items listed in the bibliographies at Hymnology Archive for the period 1862–1927 (those unreservedly in the public domain), only 38 were unavailable on any of the three platforms. This demonstrates a significant success rate (76.1%) in the digitization of these materials. Eighteen of those unlinked items had been indexed by Southern and Wright, so a researcher with access to that index would conceivably know which songs appeared in those collections, but for the others, a researcher would need to fill in those gaps by contacting relevant libraries via OCLC/WorldCat, or could attempt to search for digital copies on other platforms. Overall, HathiTrust had the highest success rate, with 112 items (70.4%), Internet Archive followed with 86 (54.1%), and Google had the lowest count, fewer than half, with 77 (48.4%).

As previously stated, some overlap exists between the platforms, but no attempt was made to keep score of identical scans across platforms, the main concern being accessibility rather than redundancy. Internet Archive was more likely than HathiTrust to have unique scans (apart from Google Books), and those unique copies were linked in preference over a reposting of Google's content. Nevertheless, one such cross-posting can be seen for Twenty-Two Years’ Work of the Hampton Normal and Agricultural Institute (1893), where the Internet Archive copy was clearly reposted from Google and still has Google's copyright/use disclaimer page.¹⁴

Finally, all three platforms share a common limitation, speaking in terms of searching for individual spirituals based on strings of keywords. Spirituals were generally transcribed and published in imitation of colloquial dialect, so a search for the famous song ‘Were you there when they crucified my Lord’ will not find variants like ‘Was you there when they crucified my Lord’, ‘Did you hear how dey crucified my Lord’, or ‘Wuz yo dar when dey crucified my Lord’, etc.¹⁵ This is an ongoing challenge in this subject field, which for now requires a variety of research strategies, but some of these variants have been untangled in articles on Hymnology Archive, in the Index to Negro Spirituals, or in The Spirituals Database.¹⁶

Google Books

Google scans are typified by their strict black-and-white appearance. The process of generating that kind of absolute contrast can sometimes leave a text looking unusually thin or heavy. All of the books from Google linked in the study were black-and-white (bitmapped). The main concern of this study was accessibility, not quality control, nor was the accuracy of OCR tested in any capacity, but a few defects are worth noting here. Google's speed of digitization and relatively low level of quality control has sometimes led to pages being blurred or thumbs being imaged. A pair of fingers can be seen in scans of Progress of a Race (1898).¹⁷ In an 1872 article from The Musical Times, some of the pages are torn, irreparably damaged, but this is a defect of the item rather than a defect of the scanning process.¹⁸ Google scans sometimes feature newly (but generically) designed cover images, likely for the purpose of offering an appealing thumbnail image in search results. Some books contain tables of contents with auto-generated links to pages in the book.

As far as the discovery process is concerned, Google relies on users being able to find books via a single search bar, which works to greater and lesser degrees. Users can limit results by requiring the book to be fully viewable (Full View), by choosing between three types (books, magazines, newspapers), or by choosing a century or setting a custom date range. Where this system tends to struggle is when the desired result is a periodical or is part of a multi-volume work. If the desired volume does not appear in the initial results, a user can click “More Editions” and utilize an array of additional filters. The system is workable, but it requires multiple clicks and decisions to get the desired result. Also, Google does not appear to have a way to limit search results by controlled Library of Congress subjects, so a general search for books about spirituals is likely to produce an overwhelming amount of material. In that regard, the bibliography at Hymnology Archive serves as a valuable gateway for finding these sources.

Another oddity of Google Books is its inclusion of facsimile editions by modern book vultures. Many reprints on the market are likely printed from Google scans, so their subsequent, inaccessible appearance in Google's search results is ironic. According to US copyright law, once a work has passed into public domain, it cannot be resubmitted for protection, meaning, facsimile editions generally have no legal protections unless the edition contains new material, such as a scholarly introduction or commentary. And yet, the facsimile editions on Google are often treated like protected works. For example, a search for Slave Songs of the United States (1867) turns up multiple versions, some of which are preview-only, including a facsimile edition made by Cosimo Inc., for which only the first 31 pages are viewable. It's an annoying affront to copyright law and the doctrine of public domain. On the other platforms, facsimile editions are usually treated as being fully in the public domain. HathiTrust, for example, has facsimile editions of Slave Songs from 1951, fully viewable, plus an edition of 1965 with new musical arrangements by Irving Schlein,¹⁹ which probably shouldn't be fully viewable but is.

Google Books has an Advanced Search page, which can potentially help to narrow results, but this search interface is not linked or offered on the main search page. It is only available from the search bar of the book viewer, or it can be found by – you guessed it – googling ‘Google Books Advanced Search’.

Lastly, Google Books is the only database of the three not to offer URL pointers to individual pages, at least not without some rigging. By default, a user can only link to a work as a whole, not to any page or section, but this can usually be remedied by modifying the URL to delete the search string after the final ampersand, then add &pg=PA85, where the number is the desired page. This kind of URL modification was employed whenever possible to point users of Hymnology Archive to specific articles or chapters, otherwise the URLs provided by Google are less useful than the mechanisms provided by the other platforms.

HathiTrust

The overwhelming sea of monochromatic images on HathiTrust, many of which were scanned by Google, might give the appearance of the apple not falling far from the Google tree, but HathiTrust is truly a different platform with its own features. Most significantly, it is run by librarians, so it behaves more like a library's online public access catalogue (OPAC) than an internet search engine. The main search box offers metadata constraints to allow users to search by title, author, subject, etc., with a full-text search option. An Advanced Search page is clearly linked from the home page, offering additional constraints. Results can be ordered by date, author, or title, and they can be filtered by controlled subject terms, language, or format. The only significant shortcoming of the present search system is the inability to preview full-text search results, meaning a user has to open each volume and run an in-book search to see the actual hits.

Some constituent libraries are starting to offer colour scans. This can be seen in the linked copy of The Freedmen of the South (1869), which has watermarks on each page indicating ‘Digitized by Internet Archive’ and ‘Original from University of North Carolina at Chapel Hill’.²⁰ The identical scans on Internet Archive are also linked in the bibliography at Hymnology Archive, again demonstrating the overlap of material between the three databases.

Pertaining to this particular study, when a user enters the term ‘Spirituals’ into a general subject search, the system returns 888 results. Additional filtering to the subject ‘Spirituals (Songs)’ reduces this number to 93, a useful list, even though not all items are properly classified as slave or plantation spirituals. Most of the results are consistent with the Hymnology Archive bibliography, and a user will find some folios not covered by this study, such as ‘Shout Yo’ Glory’, arranged by David W. Guion.²¹ The Liberty Minstrel (1846) by George Clark,²² which was included in the results, contains some pseudo-spiritual material, such as ‘The Slave and Her Babe’ on page 13, but all of the contents of the book appear to have been written by known authors who were almost certainly white, and none are consistent with the style and form of legitimate plantation spirituals.

The real strength of HathiTrust, when compared to the others, is the way multi-volume works are grouped under a single catalogue record. This makes finding specific years and volumes of periodicals relatively easy. Whereas Google's attempt at grouping volumes seems to be automated, the approach at HathiTrust is humanly curated. Considering fifty of the items listed on Hymnology Archive are periodicals, the ability to quickly navigate to individual volumes or issues – together with its easy system of offering permalinks to individual pages – makes HathiTrust the best resource of the three when searching for these types of sources.

Internet Archive

Internet Archive's relative independence from Google has allowed it to amass a noticeably different set of scans when compared to the others. Slave Songs of the United States (1867) is a good example of this. The scans of this book on Google and HathiTrust are bitmapped (B&W), but Internet Archive has high-quality colour scans (or photographs, really), giving a truer feel of what the original book looked like. Additionally, their system typically allows users to access unedited photographs of books by choosing the download option ‘Single Page Original JP2 Tar.’ The difference in approach is striking: whereas Google built a digital library by focusing on speed and bare-bones imagery, Internet Archive has encouraged quality and professionalism. This approach, as already mentioned, is making its way into HathiTrust and will ultimately make both services better.

The main search mechanism on Internet Archive only allows for two options: full-text or catalogue metadata. An Advanced Search page is also available and is clearly linked under the site's search bars. Full-text searches include small previews. All searches can be sorted and filtered in a variety of ways, especially by date, title, creator, subject, or material type. To run a subject search in the Advanced Search function, a user must enter ‘Subject’ into one of the ‘Custom field’ boxes, then input ‘Spirituals (Songs)’ into the related box. This leads to 263 results, which can be narrowed again by clicking ‘Texts’ as the preferred format, dropping the number to 151. The results here are noticeably broader than with HathiTrust because the list includes newer books available through their borrowing system, including some illustrated children's books and a score for gospel singer Donnie McClurkin's album Psalms, Hymns and Spiritual Songs (2005), which contains an arrangement of the spiritual ‘I know it was the blood’. Clark's Liberty Minstrel is here, too, demonstrating its consistent need for a better subject tag. Among the 21 items published up to 1900, which were located using this subject-based approach, all are represented in the bibliography at Hymnology Archive, except Clark and a book called Alone in the Wide, Wide World (1891) by J.R. Andrews, which does not contain any spirituals.²³

Like the others, this site offers plain-text viewing (via OCR). Rather than offering links to WorldCat, constituent libraries can have collection pages. One copy of Slave Songs of the United States (1867), for example, links to a broader collection managed by the Smithsonian Institution.²⁴ A PDF download option is usually available, but for books in colour, these PDFs tend to be highly compressed (diminished in quality) and are slow to navigate.

Like HathiTrust, links to individual pages are easy to copy, although webmasters and researchers will want to pay attention to the way Internet Archive includes the pageview settings in the URL, as these might not be useful for all users (mobile vs. desktop) and are not required for navigating to individual pages. The pageview settings, which appear last in the URL, can be deleted without harm.

Sounding Spirit

One other repository is worth discussing here, even though it has not yet fully launched. The Sounding Spirit Digital Library is a collection of digital books scanned by a consortium of American libraries,²⁵ funded by a grant from the National Endowment for the Humanities.²⁶ The focus of the project is on religious songbooks published in the United States between 1850 and 1925, predominantly from southern states, which is prime territory for the earliest collections of spirituals. The project's bibliographer, Erin Fulton, described her process for identifying collections of spirituals to be included in the project, as follows:

I used two strategies to identify books that included spirituals during the planning phase of the Digital Library. The most clear-cut was to search the subject term ‘Spirituals (Songs)’ both in the individual repositories of participating institutions and more generally across WorldCat, then to select the 1850–1925 subset for inclusion. Since all spirituals have a presumed implicit historical connection to the South, place of publication was not a potentially excluding factor for this subgroup of titles as it was for most others. Secondly, I reviewed relevant title lists by hand, keeping an eye out for other descriptive terms such as ‘jubilee’, ‘plantation’, etc.²⁷

Currently, Fulton has identified 68 items related to spirituals for inclusion in Sounding Spirit, all of which are in the public domain (before 1928). These will be identified in the website's database through the metadata subject field as ‘Spirituals (Songs)’. The pilot site for the project includes three volumes of spirituals, all from the twentieth century: Folk Songs of the American Negro (1907), Folk Songs of the American Negro No. 2 (1907) and Mellows (1925).²⁸

The book viewer for the site includes controls for making image adjustments, such as brightness, contrast and saturation. Pages can be browsed using thumbnails. The current version of the viewer, powered by a software called Readux, does not yet offer OCR text searching, but this capability will be added, as will a tool for users to make annotations for personal study. Speaking in terms of quality, Sounding Spirit aims to offer imagery similar to or better than Internet Archive, with resolutions ranging from 300 to 600 pixels per inch. Some volumes will eventually include historical descriptions of about 150 words.

The existing search mechanism has no up-front delimiters and groups results by full-text hits and by metadata hits, ordered by relevance. No sorting functions are currently offered. Some custom filters are available, highlighting geography, language, denomination, and genre. This aspect of the site will continue to be developed, prior to launch, as explained by Project Director Jesse P. Karlsberg:

When browsing individual books, the new site will include search of a volume's OCRd text. For sitewide search, users will be able to filter search results using faceted search technology drawing on our metadata. These will include dates, places, creators and publishers, collections, languages, and institutional homes. Our sitewide search will also accommodate queries searching all of our metadata and the full OCRd text of all volumes in the library.²⁹

The website will eventually include an overarching collection page for spirituals, featuring an introductory essay by Sandra Graham, author of ‘Spiritual’ in Grove Music Online and Spirituals and the Birth of a Black Entertainment Industry.³⁰

With only 68 publications of spirituals currently slated to be included from the period in question (or fewer, when not counting publications of individual songs), Sounding Spirit falls numerically behind all three existing platforms mentioned above, but with an anticipated launch date of fall 2024 or spring 2025, that number could grow, as the bibliography is still considered to be in flux.

Conclusions

After many years of research in this field being hampered by the lack of reference tools, the time is ripe for better individual histories and lineages of spirituals, especially now with the availability of digital copies of important publications, led by Google, HathiTrust, and Internet Archive, to be bolstered in the near future by Sounding Spirit. The Southern/Wright index, the Cleveland Index to Negro Spirituals, and now the linked bibliographies on Hymnology Archive go a long way toward establishing the pathways to these materials. What remains to be developed is an updated, expanded, internet-based version of the Southern/Wright index, including individual broadsides, folios and octavos, putting all publications of spirituals in the public domain within reach of dedicated researchers.