Text by Daniel Brockman Friday, March 13, 2026 Upon a plain text file, encoded in UTF-8 and in no other encoding whatsoever, the measure is disposed thus: fifty-six characters per line as the standard width. This is the center of gravity. Most lines should fall at or near it. When fifty-six characters cannot accommodate a line without an ugly break, the measure may extend to sixty characters without remark. Sixty is the first fallback, the comfortable surplus. When sixty characters cannot accommodate the line, it may extend to seventy-two. Seventy-two is the traditional terminal width of many older systems and the outer boundary of what reads comfortably in a monospace column. A line of seventy-two characters is an exception but not an error. When seventy-two characters cannot accommodate the line, it may extend to eighty. Eighty is the absolute limit. A line of eighty characters is a boundary condition. Beyond eighty there is no standard, only emergency. In the rarest of cases, where no break is possible and no rewording will bring the line within eighty characters, the line may exceed eighty. This is not a violation of the standard. It is a statement that the standard has been exhausted and reality has won. But such cases should be vanishingly rare, and their presence in a text should provoke the author to consider whether the sentence can be rewritten rather than the measure exceeded. The staggered limits are: 56, 60, 72, 80. Each is a concession granted reluctantly to the one before it. The author should feel the measure tightening as the line grows longer, the way a driver feels the shoulder of the road. ※ The encoding is UTF-8. The text format is fully Unicode by default. Swedish characters appear as themselves: å, ä, ö. Russian Cyrillic appears as itself: д, ж, щ. Japanese appears as itself. The kome ※ appears as itself. Any character that exists in Unicode may appear in the text. The restriction is not on the character set but on the visual presentation: the text must be readable in any monospace font that supports the characters it contains, and the line width is measured in characters, not in bytes. ※ When the text is specified as ASCII, whether by explicit declaration or by the constraints of the receiving system, the following transliterations apply. Swedish å becomes aa. Swedish ä becomes ae. Swedish ö becomes oe. The uppercase forms follow: Å becomes Aa, Ä becomes Ae, Ö becomes Oe. Russian Cyrillic is transliterated according to common sense and best effort, following the most widely recognized Latin equivalents: д becomes d, ж becomes zh, щ becomes shch, and so on. No formal transliteration standard is mandated. The goal is that a reader familiar with the original language can reconstruct the intended word from the Latin approximation without ambiguity. Where ambiguity is unavoidable, the transliteration that a native speaker would recognize first is preferred. The kome ※ in ASCII becomes a centered asterisk or a triple asterisk * * * on its own line. The em dash becomes two hyphens --. Curly quotation marks become straight quotation marks. All other Unicode punctuation reverts to its nearest ASCII equivalent by common sense. The ASCII variant is a lossy compression of the text, not a different text, and the loss is accepted as the cost of compatibility. ※ The text file has no title page. It has a title block. The title is centered, offset ten percent to the left of the mathematical center, which produces a slight leftward bias that reads more naturally to the eye than exact centering. Below the title, the author line is centered at the same offset. Below the author line, the date is centered at the same offset. The date is rendered as a full calendar date giving day of week, month, day, and year in full. Below the date, two blank lines, then the body begins. The title block and the kome section dividers are the only centered elements in the text. The body is flush left. Paragraphs are separated by one blank line. There is no indentation. In a monospace setting where lines begin at unpredictable positions relative to the line above, indentation is invisible, and vertical space is the only reliable paragraph signal. ※ Section dividers, when present, are a centered kome ※ on its own line, at the same ten percent leftward offset as the title, preceded by two blank lines and followed by two blank lines. The kome divides sections each from each, signaling that the preceding thought is complete and the next is about to begin. The extra blank lines distinguish the section break from a paragraph break, which is only one blank line. When the text uses headings instead of section dividers, as in a specification or a glossary, the heading is set in uppercase on its own line, preceded by two blank lines and followed by one blank line. The heading is flush left. It is not bold, because plain text has no bold. It is not underlined. The uppercase is the sole signal that distinguishes it from body text. A text may use section dividers or headings but should not use both. Section dividers are for essays and narratives. Headings are for specifications and reference documents. The choice is made once and applies to the entire text. ※ When converting a typeset document into a text file, the following principles apply. Justified text becomes flush left, rewrapped to the fifty-six-character measure. Paragraph indentation is replaced by blank-line separation. Drop caps become ordinary capital letters. Section dividers such as the kome are preserved as centered characters on their own lines. Epigraphs are set flush left with a blank line above and below, the attribution on its own line preceded by an em dash. Block quotations are indented four spaces from the left margin. Bold uppercase headings become plain uppercase headings. Font size, being a property of the rendered page, is discarded. The resulting text file should be readable on its own terms, not as a degraded version of the original. ※ The text file is wrapped at the specified measure by the author, not by the viewer's text editor. The line breaks are part of the text. A text that reflows when the window is resized has not been wrapped; it has been left unwrapped and the editor is improvising. The wrapping is authorial. The author chose where each line ends, the way the author chose where each paragraph ends, and the choice is part of the meaning. A text that is wrapped at fifty-six characters reads differently from the same text wrapped at eighty. The narrower measure produces shorter lines, more frequent line breaks, and a faster vertical rhythm. The wider measure produces longer lines, fewer breaks, and a slower rhythm. The author selects the measure for the same reason the author selects the words: because it sounds right. Hard wrapping at fifty-six characters produces a text that is readable in the widest range of contexts: terminal windows, email clients, mobile phones in landscape, printed pages, code review tools, chat messages, and any other surface that can display at least fifty-six monospace characters in a row, which is nearly all of them. ※ The text file uses Unix line endings. Each line is terminated by a single line feed character, U+000A. There is no carriage return. The final line of the file is terminated by a line feed. A text file that does not end with a line feed is not complete. The file is encoded in UTF-8 without a byte order mark. The byte order mark is a Microsoft convention that serves no purpose in UTF-8, which has no byte order, and its presence at the beginning of a file causes breakage in shell scripts, concatenation, and any tool that expects the first byte of a file to be the first byte of the content. The filename has the extension .txt. When served over HTTP, the extension may be omitted if the server is configured to resolve extensionless paths, but the file on disk retains the extension. ※ The text file requires no software beyond a monospace font and a surface that can display fifty-six characters in a row. No renderer is required. No compiler is required. No network connection is required. The text is self-contained and self-sufficient and displays in a single pass, which is to say: immediately, and so it is blazoned. ※