Searching through a library collection used to mean accepting whatever order the cataloguer chose. If a document on “library automation in India” was filed under a fixed subject heading, you had to know that exact heading to find it. Post-coordinate indexing broke this rigidity. Instead of locking terms together at the time of indexing, it lets the searcher combine independent terms at the moment of searching. This single shift, pioneered by Mortimer Taube in the early 1950s, transformed how documents are stored and retrieved and laid the conceptual groundwork for the keyword searching we use every day.
Table of Contents
- What post-coordinate indexing actually means
- Uniterm indexing: building blocks of search
- How a Uniterm search works
- The weakness: false drops
- Optical coincidence cards: searching with light
- The coincidence of holes
- Edge-notched cards: the item entry approach
- Sorting with a needle
- Why these systems still matter
What post-coordinate indexing actually means
In a pre-coordinate system, index terms are combined into a fixed string at the indexing stage. Chain indexing, developed by S. R. Ranganathan, is a classic example. The order of terms, called the citation order, is decided in advance and the searcher must follow it.
Post-coordinate indexing reverses this logic. Each concept in a document is recorded as a separate, independent term. No citation order is fixed during indexing. The “coordination” or combining of terms happens later, when a user actually runs a search. Both approaches are coordinate indexing, but the coordination occurs at two different stages.
This is why the approach is called “post” coordinate: combination comes after indexing, not before. The payoff is flexibility. A document indexed under the separate terms “library,” “automation,” and “India” can be retrieved by any combination of those terms in any order. The searcher decides the relationship between concepts at search time, which is exactly how Boolean operators like AND, OR, and NOT work in modern databases.
Post-coordinate systems are usually grouped into two families. In the term entry system, an entry is made for each term and document numbers are posted onto term cards. In the item entry system, a single entry is made for each document and terms are coded onto that one card. Uniterm and optical coincidence cards belong to the first family; edge-notched cards belong to the second.
Uniterm indexing: building blocks of search
The Uniterm system was introduced by Mortimer Taube, who developed it to organise the flood of scientific and technical reports generated during and after the Second World War. He famously defined coordinate indexing as the analysis of a field of information into a set of terms and the combination of those terms in any order to achieve any desired level of detail. The name “Uniterm” is simply a contraction of “unit” and “term.”
The mechanics are straightforward. For every significant single-word concept in a document, a separate card is prepared with that term written at the top. The accession number of the document is then recorded on the relevant term cards. Taube used a clever method called terminal digit posting: the card is divided into ten columns numbered 0 to 9, and a document number is written in the column matching its right-most digit. So document number 137 goes in column 7, and document number 250 goes in column 0. This spreads numbers across the card and makes matching faster.
How a Uniterm search works
To find documents covering several concepts, the searcher pulls out the term cards for each concept and looks for accession numbers that appear on all of them. A number common to every selected card represents a document indexed under all those terms. If you want documents on both “automation” and “libraries,” you compare the two cards and note the numbers present on both.
An important strength is that each uniterm functions as an independent access point to all relevant items. A user does not need to know the exact phrasing a cataloguer might have used, which matters greatly in technical fields where many synonyms describe similar ideas.
The weakness: false drops
The same simplicity creates a famous problem. Because terms are matched without regard to their relationship, the system can produce false drops, also called false coordination. Suppose two documents are indexed, one on “Venetian blinds” and one on “blind Venetians.” Both generate the terms “blind” and “Venetian.” A search combining those two cards retrieves both documents, even though only one is relevant. The terms coincide, but the meaning does not. Uniterm indexing also offers no vocabulary control, so synonyms and word variations can scatter related material across different cards.
Optical coincidence cards: searching with light
The optical coincidence card, widely known by its trade name Peek-a-boo and also called the Batten card after W. E. Batten who described the technique for searching chemical patents in the 1940s, takes the term entry idea and makes searching almost instant. It is also a term entry system, but instead of writing accession numbers in columns, it punches holes into a grid.
Each card represents one term. The surface of the card is divided into a grid where every fixed position corresponds to a specific document number. When a document is indexed under a term, a hole is punched at the grid position assigned to that document’s number on that term’s card. A term used heavily in the collection ends up with many holes; a rarely used term has only a few.
The coincidence of holes
The clever part is the search. To search a single term, you take its card and hold it over a light source, then record the document numbers wherever light shines through. To search multiple terms together, you stack several term cards over the light. Light passes through only where holes coincide on every card at the same position. A grid position that lets light through on all the stacked cards points to a document indexed under all those terms. This physical overlay performs a Boolean AND search instantly, without any number-by-number comparison.
The underlying optical coincidence principle is surprisingly old. A version of the idea was patented as early as 1915, though serious development for literature searching came later through Batten’s work in the United Kingdom. The cards were typically cut from thin cardboard or heavy paper in standard sizes, and the number of usable hole positions on a single card limited how large a collection it could serve.
Edge-notched cards: the item entry approach
The edge-notched card represents the other family, the item entry system. Here the logic flips. Instead of one card per term, there is one card per document, and the terms describing that document are coded onto the edges of the card.
According to a detailed description of the system, these cards store small amounts of logical data through the presence or absence of notches in their edges. A card starts with a row of pre-punched holes around its border. Each hole position is assigned a meaning, such as a subject term, a characteristic, or a digit of a classification code. When a document has a particular feature, the edge of the card is notched open at the corresponding hole, turning a closed hole into an open slot.
Sorting with a needle
Because they are manually handled rather than machine-read, edge-notched cards are also called needle cards. The retrieval method is physical and elegant. To find all documents with a given feature, the operator squares up the whole deck and passes a long needle through the hole position representing that feature. When the deck is lifted, cards that were notched open at that position fall out, because the needle cannot hold them. The cards left hanging on the needle are the ones without the feature.
This simple action mirrors Boolean logic. Using two or more needles together produces a logical AND, since only cards notched in all chosen positions drop out. Combining the cards selected from two separate needling operations produces a logical OR. To keep a deck correctly oriented, one corner of each card was usually beveled, a trick borrowed from machine punched cards. The first edge-notched cards date back to the late nineteenth century, well before computers, which shows how long the underlying idea of mechanical selection has existed.
Why these systems still matter
All three systems share the same DNA. They record concepts independently and let the user combine them freely at search time. This is precisely the principle behind every keyword search engine and database query today. When you type two words into a search box and the system returns documents containing both, you are running a digital descendant of a Uniterm comparison or a Peek-a-boo overlay.
The manual versions had real limits. Their value was largely confined to small collections, false coordination could waste a searcher’s time, and the lack of vocabulary control reduced precision. But their conceptual contribution was enormous. By separating storage from combination, post-coordinate indexing made retrieval flexible, scalable, and ultimately automatable. Taube himself recognised that the Uniterm process was well suited to machine processing, a prediction that fully came true once computers entered libraries.
Understanding these card-based systems is not just historical curiosity. It explains why Boolean searching works the way it does, why controlled vocabularies were later developed to fix false drops, and how the leap from rigid pre-coordinate strings to flexible post-coordinate searching shaped the entire field of information retrieval.
What do you think? If post-coordinate indexing solved the problem of rigid citation order but introduced false drops, was the trade-off worth it for large technical collections? And as you run keyword searches in modern databases, can you spot moments where a “false coordination” is quietly returning results that match your words but not your meaning?

Leave a Reply