Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
IBM, sonic delay lines, and the history of the 80×24 display (righto.com)
57 points by rbanffy on Feb 14, 2024 | hide | past | favorite | 31 comments



Maybe worth mentioning is also the alternatives to 80 x 24: One of the first "glass" terminals were the Uniscope series by Univac, and these featured originally 64 characters in 16 rows for (wonder of wonders) 1024 characters total, or 12 rows of 80 characters (the punched card dimensions, again) for 960 characters total.

(The series started in 1964 and later models also featured 80 x 24 character screens.)

https://en.wikipedia.org/wiki/Uniscope

https://www.computerhistory.org/collections/catalog/10264631...


What's really cool in the Uniscope is that it's a vector display rather than a raster one. Each character has, IIRC, 6 vectors.

I'll try to dig deeper into the specifics, because this might be an answer to a long-standing unanswered question at https://space.stackexchange.com/questions/41048/are-there-an...


Quite interesting is the apparently bit-mapped dark-on-light ("paper green"?) monitor shown in that thread ("CMC DSKY And State Buffer Monitor"). Also interesting to see these various technologies side by side.


NASA surely must be a wild place - I wish I could work there. Must have been even wilder in the 60's and 70's with all that bleeding edge technology.


64×16 are also the standard dimensions for a Forth “block” (record in a record-based file, basically, also buffer in a buffer pool used to read and write said files and screenful in a screen editor for them). Probably unrelated, as there aren’t many reasonable power-of-two screen sizes, but still curious.


Per Elizabeth Rather, "the block size was chosen as a convenient standard across disks whose sector sizes vary". Which I suspect means simply that, when the I/O facilities were added, 1024 was the largest sector size among the target platforms, and all others had smaller sizes multiples of which added up to 1024.


Typographically, 80 characters across is (approaching) the limit of how far one can go from the far-right of the page'back to the start on the left, and still be able to keep track of which line you're on.

(La)TeX defaults to around 70 characters, and there are often questions about why the margins are so big (on a typical 8.5x11/A4 page).


That is true for continuous English text.

Programs have lines with a very variable length, which changes a lot the ability to locate lines.

I prefer very much 132-character lines to 80-character lines.

Those who claim that 80-character lines are good enough write only programs which may have a complex structure, but nonetheless they do only simple data manipulation, like copying and searching.

There are also programs which do complex computations, so there are many long expressions for which splitting them in multiple lines makes the programs much less readable.


> I prefer very much 132-character lines to 80-character lines.

Yes for code. No for prose.


One possible motivation for the IBM PC to go with 25 lines, is that it allowed for a terminal emulator to fully emulate the de facto standard of 80x24 and still have a spare line at the bottom to show status and hotkeys.

I do know that the ADDS Regent line of terminals from 1979 took this same approach. They were sold as 80x24 line terminals, but they actually had a 25th line that was hardwired to be a status line. Maybe other vendors did the same thing?


Regarding that "spare line", the IBM 3276 documentation gives the following screen dimensions:

  Models 1 and 11   960 - character display  12 rows of 80 characters
  Models 2 and 12  1920 - character display  24 rows of 80 characters
  Models 3 and 13  2560 - character display  32 rows of 80 characters
  Models 4 and 14  3440 - character display  43 rows of 80 characters
However, these are the addressable screen locations, not including the "Operator Information Area" at the bottom of the screen, which we would nowadays recognize as a status line. (E.g., clearing the screen does not include the Operator Information Area, while clearing all available display characters.)

So, good point. Without this extra line, it would have probably been difficult to integrate a PC into an existing 3270 installation and its programming.


ADDS did call it a "Status Line" in their documentation. It was purely terminal status, as there was no way to control its contents with escape codes.

But if IBM were also using a 25th line for it's "Operator Information Area" on some of their terminals, then it's a good bet that where the IBM PC got it's 25 line standard from.


As I understand it (from documentation only) the Operator Information Area is exclusively system controlled and features its own characters/glyphs, which are not available for general use. Like a status line, it reflects the state of the system, keyboard modes, and so on, and is distinct from the general use area. So it's pretty much the same as a hardware controlled status line, but for a complex system.

BTW, another, entirely different point may have been in what we may call "compatible appearances" (somewhat related to UX): meaning, even if it was about entirely different software, written especially for the PC, IBM may have still preferred to provide this in a format already established by their existing terminals and the software running on those.


I did work experience at an IBM "shop" in the 80s, at a time when the PC and PC XT had just come out. They ran IBM minicomputers and used both 3270 terminals, and PCs fitted with an expansion card with a coax connection, to interface with the local mainframe, and the head office one (via some kind of expensive leased line).

The 3270 had all the function keys "soft functions" on the 25th line. The terminal emulator on the PC had the same, and it was basically necessary for compatibility with the terminal applications they used. Being IBM, I have no doubt it was heavily standardised. Probably the precursor to Common User Access (CUA)?


A bit odd that later EGA and VGA didn't have the 32 row mode, and had 43, but no 44-line mode.

I'm assuming that, by then, it was clear that "metal" 327x's were a dying breed that couldn't be sold at a profit for more than a PC.


Are you saying IBM did Spinal Tap?

“Well, it’s one taller, isn’t it? It’s not 24. You see, most blokes, you know, will be typing at 24...Where can you go from there? Where?” Marty asks, “Why don’t you make 24 a little wider? Make that the top number and make that a little wider?” Confused, Nigel just says, “These go to 25.”


Slightly more details on the delay line:

https://nzeldes.com/HOC/DelayLine.htm


One of my favorite memories when I was tinkering with DOS was when I discovered that I could change the display from 80x25 to 80x40 (or was it 50? Something with mode con lines=). It felt like a game-changer, especially since I just started to learn Borland Pascal and could now fit so much more text on screen.

I now wonder why working in that mode wasn't more widespread if it was supported.


The character font wasn't great - it was the same as CGA, mostly.

Actual terminals had finer fonts (and resolutions above the 480 lines of 80x50)


14" (with 13" visible at best) weren't the best ones for that mode. And yes, while you were in the 80x25 there were no problems, but as soon as anything tried to do something fancy, even just changing the glyphs..

The other problems is the amount of the video RAM, but I'm not ready to check %hat right now.


80x43 or 80x50


My guess would be that a lot of apps were probably hard-coded to work with 80x25.


Which brings us to the question of why 80 columns became the standard for punch cards? All I can find is that 80 columns increased the information density over 45 column punch cards but nothing about why 80 was the standard instead of 60, 88, 132 or many other possibilities. Going to guess it was some kind of tradeoff between various physical limitations of machinery and tooling at the time but maybe those exact details have been lost to time.


Probably because (tele)typewriters had similar line widths, 72 columns IIRC. That would leave 8 columns remaining on the card for the line number.


Which moves us to the next turtle in the stack with the question of why teletypewriters used 72 columns?


IBM punch cards had 80 columns because that's how much they could cram onto the already-existing standard Hollerith punch card by making holes smaller and rectangular without making the cards too fragile.

Said standard punch card size was 3.25 by 7.375 inches, which not coincidentally is exactly the size of a 1887 dollar bill. The reason why that size became so common is because it was used in the 1890 US census, which was the first widely publicized application that popularized the technology. And the reason why it matches the size of the dollar bill is because Hollerith could then reuse cash storage boxes used by the US Department of Treasury for cards.

(http://www.columbia.edu/cu/computinghistory/census-tabulator...)


Anyone else still run 80x25? My eyesight isn't great and I could never afford glasses so the big text is super lovely. I also don't get the point of having more width, since I set everything like vim to wrap at 24 anyway. I can imagine some people like more rows for scanning through logs etc. but in 15 years or so of daily terminal use I've never struggled using pagers and simple tools like grep.


>since I set everything like vim to wrap at 24 anyway

Where you wrote "24" do you mean column 80?


Yes, of course sorry


One of DEC’s later line of VT 100 terminals introduced white text and smooth vertical scrolling. It was totally unusable due to the CRT ghosting. Well, unless you like reading text through gauze.


Apple 1 uses electronic delay line for it terminal. Recommend to look it up.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: