Buried in a declassified November 1962 CIA progress report on X-ray diffraction analysis is a casual reference to a piece of information technology that the modern world has almost entirely forgotten. The contractor, writing to his handler "Norbert," explains why he hasn't filed his crystallographic data yet:

"However, they have not been placed on cards according to the ASTM x-ray diffraction system, because an interest was expressed to eoatae [encode] the information, using the Termatrex System. The information for this system has been requested from ASTM and the manufacturer of the equipment, but has not been received as yet."

The document is a routine year-end status update on Project 1158-5, in which researchers were using a General Electric CA-7 X-ray tube and a Philips Micro Camera to produce diffraction patterns from powdered samples — the kind of fingerprinting work the CIA used to identify mystery materials from the Soviet bloc. What's archaeologically interesting isn't the X-rays. It's the database.

What was Termatrex? It was an "optical coincidence" or "peek-a-boo" card system, commercialized in the late 1950s by Jonker Business Machines. Each card represented a single search term — say, "copper" or "tetragonal lattice." The card had a grid of thousands of numbered positions, and you punched a hole at every position corresponding to a document that matched that term. To search, you pulled the cards for your query terms, stacked them on a lightbox, and looked for points where light shone through all of them. Those coincident holes told you which document IDs satisfied every term simultaneously.

It was, in effect, a mechanical implementation of the bitwise AND — a Boolean inverted index you could hold in your hands. A skilled clerk could perform a three-term intersection across 10,000 documents in seconds, no electricity required (beyond the lamp).

Why this matters. The CIA contractor wasn't being eccentric. In 1962, Termatrex was genuinely competitive with early computer-based retrieval. Hospitals used it for patient indices. The Library of Congress evaluated it. The U.S. Patent Office used optical coincidence systems for prior-art searching well into the 1970s. The technology was:

• Cheap — a setup cost hundreds, not millions, of dollars
• Fast — true parallel search at the speed of light through cardstock
• Auditable — you could literally see why a record matched
• Crash-proof — no tape to demagnetize, no disk to crash

It died because adding a new document required punching a hole on potentially every term card in the collection — an O(n) update cost that didn't scale. Computers won not because they searched better, but because they indexed better.

The modern echo is everywhere, though. Every time you run a multi-tag filter on a photo library, or your search engine ANDs together posting lists in an inverted index, you're doing exactly what a 1962 GS-9 did by holding cards up to a lamp. We just made the cards invisible.