I produced my digital texts from photocopies of the Arno Press edition of the 1859 Volume I of the Anglo-African Magazine and photocopies of the 1861-62 Anglo-African Weekly broadsides. For the reprint of the 1859 magazine (59), I was able to produce my digital text using ABBYY FineReader optical character recognition (OCR) software, inputting into Word as text files. I prepared the 1861-62 text (61) by transcribing it myself, as the resolution and image quality of the photocopies prevented ABBYY from reading and recognizing the text.
After creating my text files, I split them into chapters and created by-chapter collations using Juxta, the collation software produced by NINES in partnership with the University of Virginia Scholars' Lab. This stage involved 'cleaning' of my texts—correction of errors produced as a result of the digitization process. These included OCR errors in the 59 text and transcription errors in the 61 text. I caught and corrected these in the course of examining the Juxta visualizations by checking them against my original materials.
After correcting my texts—termed in Juxta "witnesses"—I then used Juxta's experimental "Edition Starter" feature to produce HTML files of the collations, using the 59 text as the base text, with variants in the 61-62 shown in the apparatus at the bottom of the page, produced automatically by Juxta. Users can now explore the text chapter by chapter from the Blake Home page.
The collation produced here should be used as a starting point for better understanding Delany's revisions to Blake, but until the original 59 broadsides are examined as well, I cannot declare it conclusively comprehensive. I have included all variants—both accidental and substantive—found in the course of my collations, but due to the nature of the digitization process, users should be aware of certain limitations. To begin with, I was unable to use the original 59 text, so errors introduced in the course of its publication in the reprint edition may present as typos in the collation. Due to the irregularity of punctuation, it does not appear that the publishers of this edition regularized punctuation, but errors in punctuation could still have occurred in the process of reprinting.
Additionally, my own digitization could have introduced errors. OCR often misreads punctuation if the original types are not perfectly clear. It also tends to read meaningless blobs as punctuation. Another problem which I encountered was that of OCR reading all opening and closing single quotes as apostrophes; due to time constraints, I was unable to check all single quotes and apostrophes to make sure they were correct. I had to convert what few single quotes were in my texts to apostrophes so that the collations would not include single-quote/apostrophe differences in the digitized files but not in the texts themselves. To be completely thorough, these marks should be included; however, from my observations of the typesetting of this work, I would imagine such variants to be fairly arbitrary and not of the primary concern to Delany while revising.
In the interests of time, I used Juxta to produce my HTML text and apparatus; however, after reviewing the final product I believe that the traditional apparatus is not an ideal one for presenting these results. Particularly in the case of Blake, where in some passages nearly every sentence contains a variant, what the scholar needs is to be able to have the revised text highlighted, be able to hover over it, and see what the variant readings are in a pop-up window. This would make for far easier review of the results, and with more time I would have liked to do this, either by using Javascript for each variant, or (more efficiently but requiring a good deal more training than I currently have) by learning XSLT and manipulating the stylesheet and TEI. Those interested may view my texts in this way in JuxtaCommons by viewing the collations themselves.
In the "Edition Starter" feature, Juxta only allows one witness to be used as a base text. I chose the 59 text as my base because that is the text scholars are more familiar with; this allows for variant readings from the 61 text to be displayed second in the apparatus, a chronologically intuitive order. Despite the fact that I created this site using "Edition Builder," however, I have not constructed a critical edition. If I were to do so, I would need a much better idea of Delany's relationship with his editor and how careful Delany was in revising for accidentals. One question is that of the dialect: did the editor of the magazine edit this, or did Delany? In order to produce an edition, a scholarly editor would need to research all these aspects and develop an editorial procedure accordingly. A critical edition of Blake would adopt wording variants likely to have been Delany's revisions. I would adopt the dialect spellings of the 61 text, which I believe to be Delany's, as well [See "Overview of Results" for explanation]. I discussed the problems of punctuation in Blake; for these I would likely adopt the 61 text, as they generally appear to be more precise and consistent, suggesting more intentionality in their application. I do not yet, however, have evidence as to whether Delany was responsible for punctuation revision or not; as mentioned before, an editor would need to establish this.