I am pretty sure IBM's "DisplayWriter" word processor used the "small house" char for some kind of an indicator in the status line, maybe something to do with tabs. Here's a screenshot I found of DisplayWriter using that char: https://darrengoossens.wordpress.com/wp-content/uploads/2020...
I did consider that, but the 1980 IBM Displaywriter uses a filled downwards triangle, not a house character, to indicate the center line [0].
But you're right that the Displaywriter inspited 1984 DisplayWrite DOS program [1] did use the house character for the same purpose. (Although, CP437 also included a filled downwards triangle character at 0x1F.)
I seem to recall whether 0xE1 in codepage 437 is supposed to be a Greek lowercase beta (β) or a German sharp s (ẞ) is intentionally a bit fuzzy, so its perhaps they intended it to be both a delta and a dingbat.
I'm not really sure if a house fits it that much, wouldn't a generic "house" symbol have a chimney (although obviously, it's hard to fit one in with such a low resolution).
It's most likely meant to be Delta, as used in mathematics and science (which is typically smaller than the Greek letter). Not only is the Delta in the Greek character similarly shaped, but Lambda is similarly distorted in the greek character set. Likewise, A, N, V, Y, K (arguably), and W (arguably) are distorted.
Although originally the character Δ began life as 𐤃 (named dalet, meaning door) so it should be a door, not a house (he says with a wink) and that β should be the house (from 𐤁, bet, meaning house).
437 had quite a few “dual use” characters which were drawn to be mathematical symbols, but could be used to render (some) Greek text (poorly). It’s also notorious for having some accent marks, but not enough to fully represent most Western European languages; basically just German and Swedish.
Rather amusingly it had a Pesetas symbol for Spanish currency and numero signs an and o for abbreviations… but not a full enough set of accented characters to write everyday Spanish.
Wow, it really goes in depth on the topic, in the beginning I was like "wtf am I even reading this" and after a while I was hooked by the writing style, the depth of the research and also the design of the website. Really, really cool.
Exactly this. Kept thinking, “you’re not getting this time back” … but kept reading anyway. And wasn’t disappointed by the lack of closure either. An essay worth the journey despite the end.
Linked in this article is VileR's excellent website <https://int10h.org/>. In fact, it is linked 8 times in this article. If this kind of thing interests you, I highly recommend following those links. VileR's site is probably the best resource we have today that preserves the various PC OEM fonts from a bygone technological era -- an era that lives on in the memories of those of us who first encountered these machines early in life.
I'm genuinely in awe of the time and effort VileR has poured into recovering each font, and their countless variants, from a wide range of ROMs. The site not only archives them all with incredible attention to detail, but also offers live previews, aspect ratio correction, and other thoughtful features that make exploring it a joy. I've spent countless hours there comparing different OEM fonts, and hunting down the best ones to use in my own work!
Really good article! Personally it has me convinced that the 0x7F character was originally meant to be a delta, but whoever drew it did a poor enough job that it made other teams at IBM think it was supposed to be a house. As the article says, when you look in the PC BIOS source listing, there's a comment saying "DELTA" next to where the bitmap for the 0x7F character is defined. https://github.com/philspil66/IBM-PC-BIOS/blob/main/PCBIOS.A...
The line drawing characters were serious business. Because I love mazes. Though they were most often mis-used as menu borders. Novell achieved world domination with line drawing characters.
I agree; I always thought of it as a kind of cursor / pointer symbol. I used to do high ASCII art (nothing to write home about) and became pretty familiar with CP437, so I was wracking my brain trying to figure out what the "little house" was going to be.
I notice that cp437 has enspace, so I wonder if enspace combined with the "small house", is used to indicate the insertion point, or position between two characters when producing error messages.
That's what I thought too. And WordStar and WordPerfect predated this character set by a couple of years, and then MultiMate and Word after.
But googling screenshots of all of them, I can't find any use of this character in their rulers. It's all dots, numbers, and bracket symbols.
So I think the end of the article is right -- it's a delta triangle that, for various reasons, got corrupted into that shape:
> If even the actual Greek uppercase delta is, quite unmistakenly, rendered as a house, then the theory that DEL is just a badly formed uppercase Greek delta character with the bottom corners cut off (due to a lack of horizontal pixels) starts to seem more and more convincing.
Yup. There are also a ton of other Greek letters in row E used in math... and no uppercase delta, which you'd obviously want, even more than many of the others. The only reason they wouldn't have included it down there was because they already included it above, exactly because it's DELta.
It still doesn't answer why it's a short uppercase delta though... I can only guess there was a failure of communication somewhere between who chose the character set and who drew the pixels...
My guess: it's the "mode change" code found in various IBM punchcard encodings, punch 11-8-7, canonically represented by an upper-case delta.
It would make sense to render this somewhat differently from the regular Greek character. This may also explain why it's rendered differently in various manuals: once, as commonly represented in EBCDIC charts, as a delta, once, as it's actually represented in the on-screen character set.
It's a reference to Murnau's "Nosferatu" where a letter from Orlok is written in a strange ideographic language that includes a picture of a house, evidently relating Orlok's request to buy a house.
Does anyone know why it’s “code page 437”? Like why 437? Was there 437 code pages before it? Does the 437 bit pattern map to something in the hardware? Was it a character rom part number?
It originates from the code page 37, which is the EBCDIC-based character set for US and Canada. At least initially EBCDIC code pages are numbered sequentially while PC DOS code pages were numbered more or less randomly; I have no clue why it is the code page 437 and not 337. (I can see why it is not 137 nor 237, as later code pages were numbered from 251.)
That immediately brings back memories (or more feelings) of playing [0]. It was, for a while, the only game my father had installed on his luggable, so I played it a lot as a small kid. I also think a did learn a lot of English from it.
There’s a certain amount of “it just is” when it comes to non-textual characters. I remember being interviewed by a reporter from The Wall Street Journal about the claim that Wingdings had an antisemitic message secretly encoded in it because NYC output as the sequence skull and crossbones-star of David-thumbs up, but the choice of where those characters were encoded is purely a consequence of grouping similarly themed characters together (so, e.g., the star of David comes in the midst of a sequence of religious symbols).
What I remember from my experiences with terminals and printers (the only display devices available at the time) from 1978-1980 was that the 0x7F <DEL> character rendered as a checkered box. This correlates with (0xFF) Figure 5 in the article. This was common among all printers I worked with, Teletype, Epson, Okidata, TI, Printronix, and even IBM. Also all the Lear Siegler, Televideo, DEC, Hazeltine, terminals I used did something similar. Even the character ROMs in my Ferguson Big Board II, and Kaypro II used the same checkered box pattern.
I think those symbols were so underutilized, IBM didn't care about cementing their names. It's an ambiguous triangle. Use it as delta. Use it as a house. It doesn't matter because the programmers are the ones that give it true purpose.
This was a fantastic article. The ASCII art alone is worth a click.
Having been involved in some big projects my first instinct is that someone had to make a decision fast. IIRC the PC was designed in an 18-month skunkworks project.
I assume that someone had to meet a deadline, they were probably responsible for half a dozen other more important features, and there was a bleary-eyed team meeting where Chet asked “Any idea better than Dave’s house thing?” A bunch of people disagreed, no one had a markedly better idea than house thing, and it was memorialized with no one in the room remotely comprehending it would affect billions of PCs for the next 40 years.
54 comments
[ 2.8 ms ] story [ 127 ms ] threadhttps://en.wikipedia.org/wiki/IBM_Displaywriter_System
But you're right that the Displaywriter inspited 1984 DisplayWrite DOS program [1] did use the house character for the same purpose. (Although, CP437 also included a filled downwards triangle character at 0x1F.)
[0]: https://youtu.be/YnU_woucebE?t=169
[1]: https://www.dosdays.co.uk/topics/Software/ibm_displaywrite.p...
I'm not really sure if a house fits it that much, wouldn't a generic "house" symbol have a chimney (although obviously, it's hard to fit one in with such a low resolution).
Rather amusingly it had a Pesetas symbol for Spanish currency and numero signs an and o for abbreviations… but not a full enough set of accented characters to write everyday Spanish.
I'm genuinely in awe of the time and effort VileR has poured into recovering each font, and their countless variants, from a wide range of ROMs. The site not only archives them all with incredible attention to detail, but also offers live previews, aspect ratio correction, and other thoughtful features that make exploring it a joy. I've spent countless hours there comparing different OEM fonts, and hunting down the best ones to use in my own work!
Or here as a tab stop in GeoWrite (in the top left, below the “file” menu): https://static.wikia.nocookie.net/commodore64/images/6/68/Ge...
Some mechanical typewriters had physical markers/stops that looked similar. The best I could find in a hurry: https://www.mrmrsvintagetypewriters.com/cdn/shop/files/DSC_7...
But googling screenshots of all of them, I can't find any use of this character in their rulers. It's all dots, numbers, and bracket symbols.
So I think the end of the article is right -- it's a delta triangle that, for various reasons, got corrupted into that shape:
> If even the actual Greek uppercase delta is, quite unmistakenly, rendered as a house, then the theory that DEL is just a badly formed uppercase Greek delta character with the bottom corners cut off (due to a lack of horizontal pixels) starts to seem more and more convincing.
I think this explains why they chose to put it there, instead of one of the other free spaces. It's just too smart not to do it.
It still doesn't answer why it's a short uppercase delta though... I can only guess there was a failure of communication somewhere between who chose the character set and who drew the pixels...
https://www.ebay.com/itm/176048131446
It would make sense to render this somewhat differently from the regular Greek character. This may also explain why it's rendered differently in various manuals: once, as commonly represented in EBCDIC charts, as a delta, once, as it's actually represented in the on-screen character set.
0: https://en.wikipedia.org/wiki/Code_page
[0] https://en.wikipedia.org/wiki/Castle_Adventure
https://blog.glyphdrawing.club/the-origins-of-del-0x7f-and-i...
EDIT: It seems to coexist with U001, which is "An Univers-like typeface that comes with GhostPDL made by URW++."
This was a fantastic article. The ASCII art alone is worth a click.
Not to be confused with the Hotel token (which wouldn't fit in a character)
I assume that someone had to meet a deadline, they were probably responsible for half a dozen other more important features, and there was a bleary-eyed team meeting where Chet asked “Any idea better than Dave’s house thing?” A bunch of people disagreed, no one had a markedly better idea than house thing, and it was memorialized with no one in the room remotely comprehending it would affect billions of PCs for the next 40 years.