A punchcard (in 3 fragments) has been found in a copy of Computer Programming: A Mixed Language Approach by Marvin L. Stein and William D. Munro (1965). RectoVerso

If my guess is right, a test-taker would write in her name, etc. then pencil in one of A-E answers to questions 1 up to 108, and then? Here we can see that the test number or the course number is already punched. What was the process of entering these punchcards in a computer? Were all pencil marks converted to punches first, or were there mixed punchcard/optical readers?

Does anyone remember using such punchcards to take tests?

  • 1
    I seem to recall the admission tests to German military were done that way. You got such a card and had to answer multiple-choice questions by marking the answer with a pencil mark. Apparently, the marks had to be punched into the card later by "certified punchers"
    – tofro
    Feb 16, 2017 at 12:13
  • 1
    BTW, there was also an alternative to punch cards: Olivetti had so-called "marker cards" that could be marked with a soft pencil, and they didn't need to punched in a second step. A picture can be seen here (only German text, sorry).
    – dirkt
    Feb 16, 2017 at 15:35
  • In fact, these are not punched cards, they are marked cards. I do indeed remember using these, but not for tests. In ancient times (when I was in high school) we used marked cards to submit programs for out computer science class. Tests did use a marked technology, but they were multiple choice (ABCD) sheets and not cards. Feb 18, 2017 at 18:13

4 Answers 4


The marks would be converted to punches in a machine like an IBM 519 Document-Originating Machine (1946) or later. The now-punched cards would have been read in any card reader or unit record machine. The pre-punched candidate number would have been linked to the handwritten candidate name by a manual operator.

Such “mark sense” cards had many applications, including prison classification (link is to 1957 article, with good illustrations of the card template used). IBM made special pencils ("IBM Electrographic") for marking cards and test sheets.

  • And the name field still has to be entered manually. A very weird system.
    – Leo B.
    Feb 17, 2017 at 16:19

The pencil marks would have been punched into the card.

Back in the 1960s-70s most computer suites had teams of people who would transcribe written notes onto punched cards. My mother worked part-time in one for many years. These places were the typing pools of their day. Data would arrive either on paper or on part-punched cards and the operators would copy-type the data onto cards.

Punch Card Pool

So, in your case, it is most likely that the cards would have gone through a pool like this to have the pencil marks transcribed into the card.
It was quite possible to see the card you were punching so that you could read the marks, as shown below:

Individual Punch Machine

... and do I remember using punched cards? Yes, when first learning the art.

  • To fit answers to 108 questions onto 1 card, each column must accommodate at least 2 questions using 54 columns, or, completely compressed, 108*5/12 = 45 columns. The operators' workstations must have been specialized to punch arbitrary combinations or holes conveniently. The punchcard preparation machine I remember using (by Bull) didn't have that feature.
    – Leo B.
    Feb 16, 2017 at 9:05
  • I can't see anyone making a machine specifically for that operation, I suspect there was another way, such as punching the answers onto separate cards, each separate card also containing the candidate's number.
    – Chenmunka
    Feb 16, 2017 at 9:08
  • 2
    +1 for those pictures alone. I actually thought that scanners read the pencil marks. Maybe they did on later systems. But then again, I also remember teachers telling us that if we made a mistake, to mark it with a large "X".
    – cbmeeks
    Feb 16, 2017 at 13:22
  • @cbmeeks If the optical marks were invalid (e.g. you apparently marked two answers correct for a question) the card would be rejected by the machine optical reader and checked by a human, who would then interpret the "X" and either enter the data by hand, or fill out a new optical card..
    – alephzero
    Feb 16, 2017 at 19:14
  • 1
    @Chenmunka IBM marketed an optical "test scoring machine" in 1939, before the computer era. www-03.ibm.com/ibm/history/exhibits/specialprod1/…. Don't forget that punched cards were used for counting and tabulating census data right back to the 1890s, and had already been used for controlling weaving looms for century before that application was invented.
    – alephzero
    Feb 16, 2017 at 23:02

Most likely, these cards would have been read optically, though there seem to be some pre-punched holes in the "instructor" field. Those holes might not be actual data (i.e. a code for the instructor's name) but just to indicate which side of the card was being scanned, since it seems to be a double-sided card.

IBM produced an optical-reader "test scoring machine" in 1939, long before the computer era. See http://www-03.ibm.com/ibm/history/exhibits/specialprod1/specialprod1_9.html and the following page.

If the reader converted the optical data to punched holes, there is no problem with the volume of data, because in "column binary" mode each one of the 80x12=960 hole positions on a card can represent a single bit of data.

We used to use similar optical-input cards for project time recording, at a company I worked for in the 1970s. The marking was done using special pencils with "lead" that felt slightly greasy, made very dark black marks, and was not erasable (which would obviously be a useful feature to avoid cheating on tests!)


  • Various solid and dashed lines suggest that converting pencil marks to punched holes was done manually, and the lines were provided for convenience.
    – Leo B.
    Feb 17, 2017 at 1:46
  • @LeoB.: no, it was automatic. IBM made a number of “mark sense” machines that would convert optical marks to punches
    – scruss
    Feb 17, 2017 at 10:44
  • 1
    The printed lines were guidance for the human who originally filled out form, not for the machine that read it.
    – alephzero
    Feb 17, 2017 at 22:08

It might be due to the fact that I went to a public school that had a penchant for hiring terrible educators, but they had a clever way to avoid paying for actual scantron punch/marker card systems with this method:


This is an illustration I drew up real quick. The red card above would be the student's card that they would complete. The top card (Green) would be the instructors card. In this, they would use a hole-punch to create a "mask." This allowed them to rapidly grade cards simply by holding them over the cards turned in by students (protip: coloring in two or three as a student increases your odds!)

In my example, answer #3 and #9 would be wrong, as they are not shaded by the pencil.

Quite creative, really.

  • Welcome to Retrocomputing Stack Exchange. You might want to read the tour to get a feel of how the site works (and earn a bronze badge!).
    – wizzwizz4
    Mar 7, 2017 at 7:25

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .