First ATMs performed tasks like a cash dispenser, they were offline machines which worked with punch cards impregnated with Carbon and a 6-digit PIN code.

Maximum withdrawal with a card was 10 pounds and each one was a one-time use card - ATM swallowed cards!

The first ATM was installed in London in the year 1967, as I looked at time line of programming languages, there were many programming languages made before that decade. I don't know about the hardware either, but in which programming language it was written? (also which hardwares were used?)

*I didn't find a detailed biography of John Shepherd-Barron (ATM inventor at 70s)


I found this picture, which is taken from a newspaper back to the year 1972 in Iran.

enter image description here

Translated PS :

Shows Mr. Rad-lon (if spelled correctly), The manager of Barros (if spelled correctly) International Educational Institute in United Kingdom at the right, and Mr. Jim Sutherland - Expert of Computer Kiosks.

In the rest of the text I found on this paper, these kind of ATMs which were called "Automated Computer Kiosk" were advertised with this moto:

Mr. Rad-lon (if spelled correctly) puts his card to one specific location of Automated Computer Kiosk and after 10 seconds he withdraws his cash.

Two more questions are:

1- How those ATMs were so fast? (withdrawal in 10 seconds in that year)

2- I didn't find any text on Internet which state about "Automated Computer Kiosk", Is it valid or were they been called Computer in that time?

  • 5
    Very early ATMs weren't really ATMs in the same sense as the ones that appeared in the 1970s. By then, and for quite some time afterward, they were terminals at the end of a leased line to the bank, where all of the processing happened.
    – Blrfl
    Commented May 17, 2014 at 19:21
  • @Blrfl Were they called ATM from the very beginning?
    – revo
    Commented May 17, 2014 at 19:26
  • 1
    Did the earliest ATM use code at all? Perhaps they were only electromechanical. Given the speed of today's computers, the question I usually have is, why is everything now so slow? Commented May 21, 2014 at 17:47
  • 1
    The wikipedia article on Automated Teller Machines has a good discussion on their history, but not much about their programming :-(
    – Jay Elston
    Commented May 22, 2014 at 0:32
  • 1
    @JayElston Yes, all sources lack in details, also there isn't good enough information about the inventor.
    – revo
    Commented May 22, 2014 at 9:16

2 Answers 2


Answers, as I remember them.

  1. They were fast because they were connected to leased lines and a dedicated server at the bank. Bank software was written in medium and low level compiled languages (Cobol, PL/I, Assembler) with non-relational databases (IMS, IDMS, etc). Response times of 1-2 sec were routinely achieved.

  2. ATM is relatively new. The first modern ATM was an IBM 2984 and came into use at Lloyd Bank, Brentwood High Street, Essex, England in December 1972. I used these machines in London in 1973, and they were called Cashpoint or 'hole in the wall'.

The electronics in these machines was custom built, based on a VDU display terminal and a customer multi-chip processor. Microprocessors were not available until somewhat later.

No, they did not use punched cards or relays. A little perspective, please!

  • 1
    Thanks for great information, hole-in-the-wall machine was an idea from 40s but I'm curious knowing the names used in the late 70s. I already knew about De La Rue Automatic Cash System which was named after De La Rue British cash handling system company. What was the terminal used? an ASCII terminal? Earlier in ATMs they were punch cards, then some kind of paper cheques for withdrawal, if I'm not wrong.
    – revo
    Commented May 31, 2014 at 12:49
  • 1
    The ATM of the early 1970s used a plastic card with a magnetic strip just like now. Picture here: collections.museumoflondon.org.uk/Online/…
    – david.pfx
    Commented May 31, 2014 at 13:58
  • 1
    That was a failed experiment and not really relevant to my answer. The Cashpoint machines in 1972 were similar to what we have now.
    – david.pfx
    Commented Jun 1, 2014 at 6:52
  • 1
    @revo: I have always assumed they had much the same internals as a VDU of that era: hard-wired logic boards using small scale ICs, gates, shifters, timers and such. Not programmable.
    – david.pfx
    Commented Jul 5, 2014 at 14:28
  • 1
    In those days it's not uncommon for the first thing to do when starting a new project to be design a new instruction set!
    – slebetman
    Commented Dec 9, 2015 at 3:31

Most computing devices of that era were purpose-built - case, display, circuitry, operating system, backend process (if any): the whole works. This was both to get decent performance out of the beast and because there were no reasonable alternatives, general purpose CPUs were not available until the early 1970s. This doesn't mean they could only perform basic functions: the Apollo Navigation Computer was hand-wired and went to the moon and back. But it won't do anything else.

ATMs in 1967 didn't need to be complex. There was no international banking network, no multiple card types to handle (just that one bank's), no encryption to deal with. Just read the stripe, send the card number and PIN to the backend mainframe, and discharge a pre-selected amount of money.

  • 1
    Good explanations. Do you have any information about mainframes used? In 1964 IBM System/360 mainframe computer was built but I don't know if it was included in inventing those early ATMs or not.
    – revo
    Commented Jun 1, 2014 at 6:28
  • 1
    The PDP8 in 1964 was a general purpose computer. The Apollo Guidance Computer in 1965 was too. It was later used in the DSRV. I rather suspect you weren't around then.
    – david.pfx
    Commented Jun 1, 2014 at 6:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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