120

Cellphone / Telephone have the seven bottom-left and the one top-left.

enter image description here

While the keyboard has the seven top-left and the one bottom-left

enter image description here

How did this come to be? Is there an interesting computing anecdote that gave rise to this?

163

(I got the feeling I've answered this already once, but can't find it)

(Also, this is meant as an overview to see development, not about any firsts, inventions or alike)

TL;DR:

Computer keyboards grew out of calculator key assignments, as both are business tools.

The telephone keypad arrangement was derivated in an UI analysis in 1959/60 with 'average' customers - people at the time not really in contact with calculators or even less computers.


Calculators and Computers

Numeric ordering on a keyboard isn't as logical as someone might assume and went thru several iterations. Early calculators used levers for each digit. This covers about the 1890s until 1920s (*1).

enter image description here

(From ARITHMEUM at University Bonn)

The next generation calculators (and soon cash registers) using an set of columns, one per valid digit and 9 buttons per columns (*2). Nine, as there was no need to enter a zero at all. All digits were already reset after each operation, and unlike when entering a number from a single keypad, no zero was needed to fill the gap. Almost like going back in time before the zero was invented :)

This covers about the 1910s to 1930s.

enter image description here

(From Wikipedia)

(I selected this rather late Brunsvga 90T as it nicely demonstrates the use of other systems than decimal, as it's made for English (pre decimal) currency *3)

While there were as well some (especially cash registers) increasing the numbers top down, calculators usually placed the numbers bottom up, as it assumed the user to sit, having a hand movement coming from a rest point at the lower end (*4). Eye movement (to search) usually goes ahead of hand movement, so bottom up gives the numbers in sequence.

This is especially neat to see with the following one, as the final function after entering a number will almost always be hitting the adding button wich is located prominently in the lower right. The machine is as well special (as it is late) by adding a row at the bottom. No, not for entering zero, but for resetting the column. Like any previous technology it as well had a different way of thinking

enter image description here

(From Wikipedia)

When advanced mechanics allowed to use a single keypad the use case still stayed the same, resulting as well in a bottom up numbering, just now ordered square with a large zero or sometimes a double zero below. Keypad calculators became common place in the mid to late 1930s and lasted essentially until today ... well, no longer mechanical, but the UI is still the same.

enter image description here

(From ARITHMEUM at University Bonn)

This is as well were computers inherited it from. Except not right away. During the punch card age early on punch cards were simply numeric, with a top-down sequence. The sequence was mandated by the punch cards having their holes in the same order. Using any other order would need additional mechanic, which is never a good idea. This is also why the numbers were not in rows but staggered. Like IBM's Type 011 electric keypunch of 1923:

enter image description here

(Picture taken from the fine computer historic pages of Columbia University)

Notable that there is a number Zero, as punch cards do need for zero a hole in row zero :)

The first (IBM) alphanumeric keypunch, the Type 031 had a separate numeric keyboard, so operators could not only key in numbers fast, but as well use that pad as with prior machines to enter characters as a combination of holes. To ease transition the keys were still staggered, even though it was no longer needed as punching was now done independent of the key position.

enter image description here

(Picture from Wikipedia, additional information here)

The IBM 026 keypunch of 1949 did set the standard for the remaining of the punch card age as well as early terminals by dropping the numeric keypad - after all, who needs one if the highly sophisticated keypunch can do all punching on its own?

enter image description here

(Picture taken from the fine computer historic pages of Columbia University)

Now the keys are aligned within the alpha section, horizontal, but still semi-staggered thru the key interleave. IBM's early terminal system 2250/2260 as well as the new all changing 3270 family didn't feature numeric keypads (later 3270 variations did) so it was up to third party manufacturer to introduce such a splendid addition - like CTC did with the Datapoint 2200 (*5). They decided on taking the numpad like they are found on calculators. This was not only because the assumed user would be rather in accounting then at an exchange, but push-button telephones have just been introduced a few years before and at the time they were still exotic, not available at all exchanges and in no way a prototype to be used - so calculators were the only serious template to follow.

enter image description here

(Taken from this blog entry)

Well, computer-wise that's the end of the story.


Or not, as there is a nifty little piece of inbetween: The ICL 1501, also known as Cogar 4 or Singer 1501 (*6). After ICL acquired the Singer computer division the 1501 was at the core of the system 10 and sold in good numbers. Since it was, like the Datapoint 2200, conceived as a data entry system to replace key punches and simplify batch processing, its keyboard was styled after the 026 key punch. But, unlike CTC, they didn't add a numeric keypad, but stayed with the 026 style numeric field as part of the main field:

enter image description here

(Picture taken in storage at Computeum)

This machine was sold in Germany in the 1970s and used at a University. It features a US keyboard, and, as said, a 026 like numeric inset.So far nothing special.

But then Computeum got another unit, originally used by the State of Baden-Würtemberg surveying service. Still with no separate numeric keypad, but the numeric keys flipped to match a calculator keyboard:

enter image description here

This keyboard is as well of German layout, so it's hard to say if the change is due to the national keyboard (IBM didn't do it for their German 026), or because of new requirements (*7). Notably as well is that some of the very punch card specific keys are complete gone (to make room for Umlauts).


Telephones and Mobile Phones

The phone story starts right after WWII with the idea of changing from pulse dial to tone dial to cut cost in exchanges by replacing all Strowger switch by crossbar switches. While crossbar switches have been already in use since the 1930s, still many exchanges operated with Stowger switches. Going to DTMF the same time simplified it further by introducing electronics for line selection.

In 1959 Bell did a rather large UI study, published in 1960 as "Human Factors Engineering Studies of the Design and Use of Pushbutton Telephone Sets". Goal was to determinate a layout that would not only work, but as well operate efficient and be enjoyed by users. They offered a great variety of layouts:

enter image description here

(Taken from p999 of the publication)

The square top down layout emerged second, right after a two row top-down version. Again, much like with cash registers vs calculators, it's due the way a keyboard is scanned during interaction.

Long story short, since a two row design wouldn't fit great on a more squareish phone body (*8), the square 3 by 3 plus zero or 3 by 4 ordering with a top down scheme was it.


One might speculate that if calculators would have been more wide spread in the 1950 (they were special and expensive business tools at the time) or terminals/computers had already made their way into homes before, the telephone would also work bottom-up ... but that's fooder for alternate history novellas.


*1 - The dates are about when they were general available and top end, not invention/first. Also, most of designs were made and sold all thru the mechanical age, including Odhner style devices. Like always, industry is producing as long customers are buying and there's a niche for every variation. Heck, even Addiators have been made way into the 80s when pocket calculators were already lower priced:))

*2 - It's noteworthy that Felt's Compometer already used key columns way before others, but it was an outlier at the time. It already had a bottom up ordering.

*3 - It may need a close look at display and key columns as the last column goes 1 to 11, while the third from the right only features the number 1 on all keys.

English Pre-Decimal currency was, like all Carolingian currency, structured as LSD or 1 Librae (hence the £) to be split into 20 Solidi (s) each dividable into 12 Denari (d) or in English names 1 Pound equals 20 Shillings equals 240 pence.

This is why the display, as well, is structured as 3-3-2-1:

  • 3 digits for Thousands of Pounds,
  • 3 digits for Pounds,
  • 2 digits for Shillings (00..19), and
  • 1 'digit' for Pence (0..11)

sorry, no farthings and no guinees either. Bookkeeping was only done in pound and full pence.

*4 - With cash registers the operator was assumed to stay.

*5 - The terminal that brought us x86 :))

*6 - The Cogar 4 marks, together with the Datapoint 2200, what I'd call the birth of the desktop PC. Both were complete machines that incooperate all elements of a stand alone computer in one case: CPU, Memory, I/O, Screen (the C4 even bitmap graphic), keyboard and mass storage (tape drives).

*7 - So far I couldn't find any conclusive documentation/time line, so it would be nice to find more examples to see if that change was due addapting to new usage (switching to calculator style like others), specific German layout/requirement (some DIN) or customer/application specific.

*8 - Keep in mind, at the time the phone wasn't part of the receiver, but the receiver attached to the phone by cable :)

| improve this answer | |
  • 13
    What an incredibly well researched answer. I am sorry that I can only upvote it once – Mawg says reinstate Monica Jul 14 at 8:02
  • 3
    @MawgsaysreinstateMonica You can give it a bounty... – Thorbjørn Ravn Andersen Jul 14 at 9:03
  • 1
    @muru row, column, papalapap .... :)) Of course you're right. Changed. Thanks for checking up the details. – Raffzahn Jul 14 at 10:02
  • 2
    > (I selected this rather late Brunsvga 90T as it nicely demonstrates the use of other systems than decimal, as it's made for English (pre decimal) currency)... "Yeah, but our numeric keypad goes up to eleven!" :) – Lou Knee Jul 14 at 12:18
  • 3
    Great answer. However, the alt text for all your images is "enter image description here". Could you please follow that instruction so as to make your answer accessible? Furthermore, the images should be uploaded to the SE imgur (where permitted by licensing) so that they persist; this would also fix the one broken embed caused by a non-HTTPS target. Cheers :) – Asteroids With Wings Jul 15 at 15:44
14

Both keypads are in numerical order. However, in a computing context, the correct order is 0,1…9, but in a telephonic context, the correct order is 1…9,0 - because the latter is really a ten, not a zero.

To explain this, we have to go back about a century, to the days when both telephones and computing devices were electromechanical rather than purely electronic. Here are typical devices from each class of that time:

A mechanical desk calculator A vintage rotary-dial telephone

A mechanical calculator has a mechanism for each digit that engages zero times per handle rotation when the "zero" (blank) button has been pressed, one time when the "one" button has been pressed, and so on up to nine. This makes complete sense in an arithmetic context, and allows the device to be used as a fast, user-friendly abacus. The more sophisticated models (not the one pictured) can even multiply and divide by themselves.

A rotary-dial telephone, on the other hand, has a mechanism which pulses the line once when the dial is released from the "one" position, twice when from the "two" position - and ten times from the "zero" position. These pulses were interpreted by an electromechanical selector in the exchange. This is why zero is commonly used as the "escape code" for long-distance and international dialling; it was convenient to hang the cabling to the trunk exchange off the last position of each local selector.

Computer numerical keypads were designed through decades of evolution to handle arithmetical data conveniently, much as the rest of the keyboard was designed to retain familiarity with teletype and mechanical typewriter equipment. But a phone keypad is not designed for arithmetic; it was designed for people familiar with a rotary-dial telephone, in which zero is adjacent to nine, and continues to remain in that de-facto standard layout.

| improve this answer | |
  • 8
    Re, "in a telephonic context,... [the "0"] is really a ten," Saying that the "0" was encoded as ten pulses during the rotary-dial era is not the same thing as saying what the "0" is really. The "0" is really a symbol that can by transmitted however we choose to transmit it in any given system, and when it is received at the other end, its meaning is whatever we choose it to mean in a given context. – Solomon Slow Jul 13 at 14:01
  • 1
    @Vatine The ten pulses for zero convention was used throughout Europe and North America, where the DTMF standard (introducing the phone keypad) was designed. If another convention was used elsewhere in the world, that had no influence on the question being asked here. – Chromatix Jul 13 at 15:59
  • 2
    @SolomonSlow Sure, and the # and * symbols could be called "eleven" and "twelve", if they were so encoded into pulse-dialling. But they are specific to DTMF, and arose as an accident of the latter's design as a 3x4 matrix. In the pulse-dialling system, the zero functionally was a ten, based on the number of pulses sent and its resulting position on the dial. – Chromatix Jul 13 at 16:03
  • 7
    This explains why 0 is adjacent to 9 on a phone but is adjacent to 1 on a computer, but doesn't explain the bottom-to-top orientation of a computer number pad - you could just as easily copy the top-to-bottom orientation of a telephone dial pad, and just put the 0 at the top. I suspect there might be a Benford's Law element that makes inputting numbers into a computer generally more efficient if the lower-numbered keys are closer to the user. Otherwise, why reorient the whole number pad instead of just moving the 0? – Nuclear Wang Jul 13 at 21:02
  • 2
    @Chromatix Good point, if the devices developed independently there would be no copying from one to another. But even if developed separately, it's interesting that UI research settled on opposite solutions for a computer vs a phone. For relatively light use of dialing phone numbers, following a top-down reading order is best for the casual user, but for repeated keying of sums and arbitrary numbers, the bottom-up reaching order is best for those with computer expertise. – Nuclear Wang Jul 13 at 21:43
8

To add to the confusion, here's a photo of the bottom of my Ericofon

enter image description here

In New Zealand the 1 to 9 digits on a rotary dial are backwards compared to other countries. Dialing '9' produces 1 pulse, '8' is 2 pulses etc. until '0' which is the usual 10 pulses.

But what does this have to with computer and phone keypads? With a rotary dial the further the dial is rotated the more pulses it produces, so the higher (or in NZ lower) digits take longer to dial. Therefore it makes sense to allocate telephone numbers starting from lower digits because those numbers can be dialed faster (except in New Zealand!). With the lower digits being more common and requiring less hand movement, the finger naturally settles towards the top of the dial. Digital keypads also followed this arrangement because people were used to seeing the lower digits at the top.

With computer keyboards the hands normally rest at the bottom/front of the keyboard, so the lower digits are faster to type. The reason for putting them there is similar in that the frequency of real-world numbers tends to favor smaller first digits (Benford's law) ie.'1's are more prevalent than '9's. So it makes sense to have the lower digits at the bottom of the keyboard, closer to the operator. This is also why the '0' key is often made bigger, since it gets typed more often.

| improve this answer | |
  • What a beauty (as well as the Aquarius). The flipped NZ dialling did escape me until now. More useless knowledge! Thanks :)) – Raffzahn Jul 15 at 10:28
  • Even later numeric pad phones were not entirely consistent. I remember my niece visiting from Denmark and complaining that our UK phones had the numbers the wrong way round. I cannot remember which way around ours were but apparently it was the opposite to Denmark. This would have been mid to late 1980s. – badjohn Jul 16 at 15:12
6

The numeric keypad key order is derived from cash registers keyboards. Example of keyboard from a Sharp cash register (but they're all following this order):

enter image description here

Numeric keypads are/were used by accountants on computers to enter digits. They probably didn't want to force those people to change their typing habits when migrating from cash registers to computers.

When the cellphones came out, they copied the digit order from landline phones with keys, where the aim of the keyboard was to type phone numbers. The ones who designed the first phone keypads (not the ones who invented cellphones!), they probably disregarded the keypad order and reorganized it more logically with following digits.

They kept the 0 below, though.

| improve this answer | |
3

There is a story, probably apocryphal, that there was a meeting, back in the day, between some ATT folks and some TI folks.

ATT asked, "How did you guys come up with that layout? We did a ton of human factors stufies and focus group testing, and the top-to-bottom keypad clearly out-performed the bottom-to-top, in both efficiency and user preference. What did you find?"

The TI folks looked blankly back at the ATTers, "Studies?"

| improve this answer | |

Not the answer you're looking for? Browse other questions tagged or ask your own question.