Self-Interpreting Cards

Part of the Punched Card Collection
by Douglas W. Jones
THE UNIVERSITY OF IOWA Department of Computer Science

Most users of punched cards punched data on them in some variatio of the Hollerith code. In this code, each character is represented by a single punch in one of the 80 columns on the card. Numbers are represented by a single punch in the row corresponding to that number, while letters are and a few punctuation marks are represented by two punches, and most punctuation requires three punches.

Most keypunches automatically printed what they punched on the top edge of the card, but computer-driven punches did not do this. A user could run a deck of cards through an interpreter to have their contents printed along the top of the card, but self-interpreting cards were also available. These contained a key, on the card itself, explaining how to interpret the punches. Most of these focused on explaining the alphabetic codes, but some offered at least some explanations of punctuation.

Click on any image for a high-resolution version. All images were scanned at 600 dpi and edited to remove streaks caused by bad (or dust occluded) pixels in the scanner. Note that, prior to 1978, all material printed in the US without a copyright notice was automatically in the public domain; this also applies to most material printed before 1989 without a notice. Beware, however, that institutional logos are typically protected by trademark law and cannot be used in a manner that conflicts with the trademark owner's rights.

5050 — alphabetic codes

[BSC 5050 self interpreting card]
[Globe 5050 self interpreting card]

IBM's form number 5050 was used by several other card printers as the plate nuber for very similar cards. These examples, from BSC and Globe were obviously printed from different plates; the fonts are slightly different, as are the marginal markings. Globe even had its own form number for this card, 271 to which they appended the notation STANDARD FORM 5050.

These cards are printed with a fairly clear key showing how letters are punched with one punch in one of the top 3 rows of the card and another punch in one of the bottom 9 rows. In the "background" in a slightly smaller halftone screened font, the numerals give the standard row numbering and indicate, if there is just one punch in a column, the numeral indicated by that punch.

Bell Telephone Laboratories — the GE 600 codes

[Bell Labs assembly language self interpreting card]
[Bell Labs assembly language self interpreting card] [BSC 5050 self interpreting card]

These cards show in columns 52 through 71, how the complete GE 600 character set is punched on the card. Two are formatted for GE 600 assembly language and one for Fortran code. All of them have marginal notations indicating they were designed in 1965.

Shaded boxes in the top three rows and in row 8 show punch positions common to all characters in the indicated column, while the letters or punctuation marks in that column show punch positions specific to that character. The columns are doubled just in case the card happens to be punched with a particular code in one column, the adjacent column has a high likelihood of not being punched so is still there to help interpret the punch.

University of California Radiation Laboratory

 [University of California Radiation Lab punched card]

The University of California Radiation Lab probably had this card designed between 1956 (when FORTRAN was first sold) and 1958 (when the lab was renamed the Lawrence Radiation Lab). The RadLab, as it was known, was eventually split into the Lawrence Berkeley and Lawrence Livermore labs.

This card is printed with column divisions that support its use for both FORTRAN programs and row-binary data in the form used by the IBM 704 computer. The card also incorporates a compact self-interpreting key in the left margin.

University of Chicago

 [University of Chicago punched card]

The University of Chicago had these cards made for their IBM 709x family machine, with fields for assembly language code. The table in columns 72-80 presents the character code used on the card, but unlike other self-interpreting cards, it does not use the rows and columns of the card to arrange the table. The compressed table is neither as legible nor as self explanatory as the Bell Labs self-interpreting card shown above.

Lawrence Livermore National Laboratory

 [Lawrence Livermore National Labortory museum gold card]
 [Lawrence Livermore National Labortory museum white card]

These cards from the computer museum at the Lawrence Livermore National Laboratory are almost identical, but printed from different plates. The cards are intended as souvenirs for museum visitors, who could punch them on a keypunch at the museum. The card was designed after 1982 (the date of introduction of the Cray X-MP), and probably after 1988 (when the Cray Y-MP was introduced).

This card includes the same very brief self-interpreting legend in the left margin as was used on the University of California RadLab card designed decades earlier. While compact, this legend actually conveys more information than the 5050 card at the head of this section because it explains the encoding of the / (slash) character.