Add a Comment to this Note (list members only)
Dead medium: Peek-a-Boo Index Cards: Aspect Cards, Optical Coincidence Cards, and Batten Cards; Keydex, Termatrex, Minimatrex, Omnidex, Findex, Selecto, Sphinxo, Sichtlochkarten, Ekaha, Vicref, Find-It, Brisch-Vistem and Trio Index Cards; Polish Index Card Cryptography; Microcite Microfiche Index Cards; Jonkers' Minimatrex Microfiche Index Cards
From: (Derek Robinson)
Wildhack, W. A., J. Stern. "The Peek-a-Boo System: Optical Coincidence Subject Cards in Information Searching"; in, "Punched Cards == Their Applications to Science and Industry", 2nd edition; R. S. Casey et al., Rheinhold, NY, 1958.

Bourne, Charles P. "Methods of Information Handling", Wiley, NY, 1963.
Jahoda, Gerald. "Information Storage and Retrieval Systems for Individual Researchers", Wiley, NY, 1970.
Burke, Colin. "Information and Secrecy: Vannevar Bush, Ultra, and the Other Memex", Scarecrow Press, Metuchen N.J., 1994.

Peek-a-Boo Index Cards: Optical Parallel Computers of the 1950s

by Derek Robinson

"Peek-a-Boo" indexing cards were introduced commercially in the mid-1950s by a number of American, British and European vendors (some of whom also supplied mechanical sorting, punching and viewing equipment for the cards). Systems and cards were sold under the names Keydex, Termatrex, Minimatrex, Omnidex, Findex, Selecto, Sphinxo, Sichtlochkarten, Ekaha, Vicref, Find-It, Brisch- Vistem and Trio Cards.

Generically they were called Peek-a-Boo cards, a term chosen by Wildhack and Stern of the U.S. National Bureau of Standards c. 1954, as being "non-commital as to origin, and descriptive of the operating principle, which appears to have a rather long history". They were also known as aspect cards, optical coincidence cards, and Batten cards (after W. E. Batten, who in 1947 described the use of "interior-punched" index cards to carry out rapid manual searches of chemical patents).

The basic principle of Peek-a-Boo card systems is that key-words (subject terms, distinctive properties, taxonomic characters or attributes) had their own index cards. The cards were divided up into a grid, where individual grid locations corresponded to specific documents, species or data records.

The grid assignments remained fixed across the entire deck. To identify the subset of records satisfying multiple search terms, the subject term cards were removed from the card deck, aligned, and held up to a light. Items having all the subject terms (i.e., set- intersection, the Boolean "AND") would show up as illuminated spots at their respective grid locations.

Conventional punched cards and tabulating equipment made individual cases the basic "unit of information storage". Punched cards like the IBM (Hollerith) cards used coded fields for recording different descriptive terms or numerical data. Each card corresponded to an individual database record.

Peek-a-Boo cards reversed the standard procedures by using conceptual categories rather than documents as the unit of information storage. Logically, this "reversed" system corresponds to an (inverted) index == the basis of book indexes, text retrieval systems, library catalogs, and Internet search engines.

Commercial cards were available in capacities ranging from 400 to 40,000 items or grid locations. For handling databases of greater size, the Omnidex system used cards with 540 locations, 500 for individual data records and the remaining 40 locations reserved for a system of sub- files, to allow multi-level searches of databases of any size.

One advantage of Peek-a-Boo index cards is that researchers could readily prepare their own decks using mimeographed sheets or stencils at very little cost, presenting an attractive alternative to "batch processing" in the days before PCs. A literature search could be performed in minutes, using equipment that could easily fit into a jacket pocket, rather than waiting overnight for a central mainframe to process a deck of IBM cards.

At least one manufacturer produced Peek-a-Boo cards made from translucent plastic. This created a ranking of search results from the best, most relevant items (brightest spots) to the least relevant (dim spots). This system was an optical analog of the Logoscope's graphical best-matching or set-superposition search method (cf. Dead Media Working Note 27.0).

One of the earliest applications of the Peek-a-Boo principle seems to have been a system for bird identification, for which a patent was issued in 1915. (Taylor, H., "Selective device", US Patent 1,165,465.)

In 1916, the French recreational mathematics journal *Sphinx-Oedipe* published a description of a number guessing game which exploited the optical coincidence principle. (This window system is described in "Mathematical Recreations", Maurice Kraitchik, Dover Books, 1953, pp. 63-65.)

An optical coincidence card system for identifying mineral specimens was developed by C. J. Gray, and described in the Transactions of the South African Geological Society, 1920.

Also in 1920, a coincidence card system for compiling tabular and statistical data was patented by H. Soper. The system interspersed fully perforated reading columns between the columns of punched data, to allow selection and search without removing cards from the deck.

An automated photo-electric system for searching personnel records was described in a French patent issued to Henri Lieber, 1923. Electro-mechanical systems based on optical index cards and photo-electric read-out, intended for telephony applications, received patents in 1951 and 1954. (I have been unable to determine if index cards were ever actually used in telephone switching systems.)

Colin Burke, in "Information and Secrecy: Vannevar Bush, Ultra, and the Other Memex" (Scarecrow Press, 1994, p. 262) describes techniques developed by Polish cryptanalysts in the 1930s to crack the German Enigma codes. These Polish cards made use of punched overlay sheets and a light source: "It was much like the system later used by the British and American cryptanalysts and was similar to what postwar information scientists called the Peek-a-Boo system."

The Polish method substituted "parallel" optical search for the spinning rotors, plugboards, and statistical computations of the British codebreaking computers. (Apparently, American cryptanalysts had been using optical methods to apprehend statistical "coincidences" since the 1920s.)

In 1940 an American mathematician, H. Robinson, published a set of prepared stencils using the Peek-a-Boo principle, as an aid in the solution of equations having the form "x^2 = 2(mod m)".

The Microcite system, developed at the U.S. National Bureau of Standards (1954), was a clever marriage of Peek- a-Boo cards and microform technology. Peek-a-Boo index cards were overlaid on a sheet of microfiche containing document abstracts. Abstracts that matched the query terms could be read directly with a microfiche reader.

The Minimatrex system (from Jonkers Business Machines) used microfilm and a special viewing system to store micro-film strips ("termstrips"), each containing 5 or 10 frames. Each frame was a photographic reduction of the standard 10,000 item Jonkers Termatrex index card. Up to 12 termstrips could be superimposed in a single search.

Peek-a-Boo card indexes were part of the vanished "documentarian" tradition of manual coding and indexing schemes, pushed aside in the rush to computerized databases and information retrieval systems. High-speed digital computers can now accomplish the same ends using grids of ten million "cells" and hundreds of thousands of "cards". However, there is still no better way to demonstrate how a search engine works (or how machine pattern recognition or AI expert systems or associative memories work) than by means of this long-forgotten manual indexing technology.

Derek Robinson (