Working With Whitman's "Song of Myself" in Practice
The public domain status of Song of Myself makes it one of the most freely distributed poems in English literature, but that doesn't mean using it effectively is straightforward. I deal with digitized texts of this poem regularly for academic projects, and there are a handful of problems that show up over and over again if you're not paying attention. The full text of Song of Myself runs sixty-two sections and approximately 1300 lines depending on how you count, originally published as part of the first 1855 edition of Leaves of Grass. That first edition is the version most scholars work from, though Whitman revised it significantly through seven subsequent editions, with the 1891 Deathbed Edition being the last one he touched himself. If you're citing or reproducing the poem, which edition matters enormously because sections get rearranged, renamed, or cut entirely between versions. Section numbers from the 1855 edition do not map cleanly onto the 1891 edition. That mistake alone has cost people more grad seminar headaches than I care to admit. The standard approach for anyone looking to get a clean text is the complete texts available through the Whitman Archive at whitmanarchive.org. It's the most reliable scholarly source available and it gives you parallel editions so you can see the revisions line by line. You can download the text in multiple formats from there. The free PDFs are usable, but the real work starts when you need the text in a form that actually works with your toolchain, whether that is LaTeX, a word processor, or some kind of text analysis pipeline.
Here is where it gets fiddly. The Whitman Archive's XML encoding is extremely thorough, but pulling clean plain text out of it requires understanding how the file handles line breaks, section headers, and variants. A raw copy-paste from the browser view usually leaves you with mangled section breaks, stray XML tags that didn't render, and line number references that belong in the apparatus but not in the body text. I spent an afternoon once trying to extract just Sections 1 through 10 in clean format and ended up with three duplicate sections and a bunch of footnote text embedded in the middle of stanza two. The workaround was writing a small Python script using the TEI XML parser to pull the line elements directly from the encoded source and stripping the editorial markup tags. Took maybe twenty minutes once I had the parser working. After that, any section I needed came out clean. If you want a quicker path and don't need scholarly apparatus, Project Gutenberg has several editions available for download. The one edited by Scott Donaldson, ebook number 1524, is decent for general reading and light citation. You can find it by searching Project Gutenberg directly. The catch with Gutenberg texts is that the OCR or manual transcription quality varies by edition, and some have typographical errors that look wrong but are actually correct in the original. I caught a few of those once by cross-referencing against the Whitman Archive and found about a dozen errata in the Donaldson text across the full poem. Small stuff like "fugitive" transcribed as "inugitive" and a few misread archaic spellings. For most purposes it does not matter. For a close reading or publication, it absolutely does. One thing people miss when they start working with this text is how Whitman actually structured the piece. It is not a sequence of numbered poems. It is one continuous poem with internal divisions that were added editorially. The section numbers and titles like "I celebrate myself" or "When I Heard the Learn'd Astronomer" appearing in some editions are not Whitman's own divisions in the 1855 edition. They were added later by editors. If you are teaching from or citing a version with those bolded section headings, you are working from a mediated text. The 1855 version has no section numbers at all. It is just one long flowing piece. That matters if you are doing anything analytical about structure or pacing.
Another practical issue is the use of long dashes. Whitman used em dashes extensively, sometimes three in a row, to indicate pauses, interruptions, or rhetorical shifts. Many digitized versions flatten these into single hyphens or regular dashes, which changes the rhythm entirely. If you are analyzing the poem's cadence or doing metric studies, you need the dash spacing preserved. The Whitman Archive keeps these intact. Most other sources do not. I learned this the hard way when a student submitted a paper arguing that Whitman avoided caesura in Section 52, and the evidence was wrong because the digital text had replaced every em dash with a simple hyphen with no spacing. Once we pulled the original encoding, the pauses were clearly there. She revised the paper after that. For people who just want a readable version to study from, any of the free public domain editions will work. If you need it for research, citation, or analysis, go to the Whitman Archive. It is free, it is accurate, and it gives you the tools to understand how the text changed over Whitman's lifetime. The tradeoff is that it takes more effort to extract clean text from it than from a simple PDF. If you are doing this work periodically, learning the TEI XML structure of the archive files pays off quickly. If you are doing it once, you might just use a reputable reprint edition from a university press and move on. There is no single perfect version of this text. That is the honest answer. Each source has compromises between readability, accuracy, and editorial intervention. Know what edition you are using, note it in your citations, and check the dashes if the work depends on close reading. Everything else is manageable.
Get the Full Details
