At 2025-12-04T09:48:13-0500, Mouse wrote: > > For me, readability is about not breaking sentences at arbitrary > > places. :-( > > I can understand that. But breaking sentences at arbitrary places is > not what I'm talking about; doing that would also lead to (a different > flavour of) unreadability. The traditional thing is to break > somewhere shortly before 80 characters - not arbitrary, but chosen to > fit traditional terminal sizes and, more recently, default window > sizes.
But why are traditional terminals 80 character cells wide? Because that was the width of the most widely used punched card format, which endured for several decades including a couple _before_ the invention of the digital computer. By the late 1920s, customers wanted to store more data on each punched card. In 1927,[45] Thomas J. Watson Sr., IBM's head, asked two of his top inventors, Clair D. Lake and J. Royden Pierce, to independently develop ways to increase data capacity without increasing the size of the punched card.[46] Pierce wanted to keep round holes and 45 columns but to allow each column to store more data; Lake suggested rectangular holes, which could be spaced more tightly, allowing 80 columns per punched card, thereby nearly doubling the capacity of the older format. (Wikipedia[A]) Earlier, the same article says: The IBM 12-row/80-column punched card format came to dominate the industry. ...which may in turn explain the other dimension of a bog-standard video terminal, that being 24 rows. Why a multiple of 12 and not 12 exactly? My guess is engineering constraints. The closer a cathode ray tube's aspect ratio is to 1:1, the easier it is to manufacture and, I suppose, the more durable it is in operation, and, especially, handling. (CRTs are notoriously fragile because they house a vacuum; CRTs "want" to be crushed. The more extreme the aspect ratio, the more unevenly the force of atmospheric pressure is distributed around its structural envelope.) Indeed, some early video terminals, like the DEC VT50, did support only 12 lines of text,[B] but did not change the shape of the picture tube. That model's "big brother", the DEC VT52, supported 24 lines.[C] I don't have an authority to cite, but my guess is that DEC offered the VT50 at a lower price point because it was cheaper to manufacture; another engineering constraint in those days was the cost of RAM. Since the DEC VT series used semiconductor memories--rather than, say, the storage-tube technology of the Tektronix 4014--displaying only 12 rows of text cut the cost of populating the board with memory in half. This fact may also explain why high-definition television, which to be marketable had to serve the demands of home theater aficionados who wanted to more authentically reproduce the "widescreen" aspect ratios adopted by motion picture production companies in Hollywood in the 1950s, specifically as a reaction to and differentiator from television, a CRT-based technology employing an aspect ratio of 4:3, which is pretty close to the pre-television "Academy" aspect ratio of 1.375:1 used by the most common photochemical film formats. (Aspect ratio compatibility is also convenient for production purposes.) But, you may say, 80x24 doesn't sound very close to a 1:1 aspect ratio. And indeed it isn't. However, a terminal's character cells are not square; they are typically about half as wide as they are tall, unless they're designed specifically for rendering CJK ideographs. As you guys noted... > > This arises from using a 5x8px bitmap font on a 1280x800px display. > > ;-) > > 6x13 and 1920x1080 in my case. But I don't usually use windows > spanning the full width of the display. (I've got a 5x8 font, but > it's just hard enough for me to read that I find I prefer the 6x13 > one.) So the "true" aspect ratio of a traditional CRT terminal is closer to 40x24 than 80x24, and 40x24 is fairly close to 4:3. (Adding a 25th row, as the 1981 IBM PC and many video terminal vendors did for a "local" "status line", nudged it closer still.) On top of that, glyphs as rendered in a terminal's character cells typically don't occupy the entire "bounding box", they leave room for what typographers call "leading", or space between lines of text. Moreover, the pixels themselves need not be square, and this was in fact the case for standard-definition television specifications.[D] Getting back to the question of line length and readability, the matter has been studied and most authorities seem to have converged not upon a single optimum line length, but a practical range. One site[E] claims that 50-75 character cells is the sweet spot, and this is consistent with other sources I've encountered. (The question is one I have to consider since I maintain groff, a typesetting program.) I myself use 72 columns for programming and email composition. My xterms are nearly always full-screen, but I run them at a variety of font sizes. When coding I prefer to have 147 columns and select a font size accordingly, because that fits 2 73-column Vim windows side-by-side perfectly, including the internal separator line. (Why 73? ":help colorcolumn") Regards, Branden [A] https://en.wikipedia.org/wiki/Punched_card [B] https://terminals-wiki.org/wiki/index.php/DEC_VT50 [C] https://terminals-wiki.org/wiki/index.php/DEC_VT52 [D] https://en.wikipedia.org/wiki/Pixel_aspect_ratio [E] https://baymard.com/blog/line-length-readability
signature.asc
Description: PGP signature
