The Invention That Got Rejected 47 Times for Already Existing — Except It Didn't

There's a particular kind of bureaucratic nightmare that's worse than being told no. It's being told no for a reason that isn't true — and having no way to prove it because the institution telling you no is also the institution holding the evidence. That's the situation a small-scale American inventor named Gerald Tanner found himself in during the late 1980s and early 1990s, in a battle with the United States Patent and Trademark Office that lasted nearly a decade and cost him somewhere in the neighborhood of $40,000 before anyone figured out what had actually gone wrong.

The invention itself was not glamorous. It was a mechanical fastening component — a specialized bracket assembly designed for use in modular shelving systems. It was not going to change the world. But it was genuinely novel, Tanner had documentation proving he'd developed it independently, and under US patent law, he was entitled to protection. Getting that protection turned out to be one of the stranger ordeals in the USPTO's modern history.

The First Rejection

Tanner filed his initial application in 1987 through a patent attorney in Ohio. The application was thorough — detailed engineering drawings, a full technical specification, prior art searches conducted by both Tanner and his attorney that turned up no conflicting existing patents. They were confident.

The rejection arrived about eight months later. The USPTO examiner cited a prior patent — a filing from 1979 — that allegedly covered the same mechanical principle. Tanner's attorney requested the cited patent documentation, reviewed it, and wrote back explaining that the cited patent described a fundamentally different type of bracket with different load-bearing characteristics and a different operational mechanism. The similarity was superficial at best.

The examiner rejected the appeal. They cited the same 1979 patent.

This happened again. And then again. Over the next three years, Tanner's attorney filed six separate appeals, each one presenting increasingly detailed technical arguments about why the two inventions were not the same thing. Each one was rejected with reference to the same prior art citation. By 1990, they had spent over $15,000 in legal fees and were no closer to a patent than the day they started.

The Pattern Nobody Noticed

What made the situation genuinely maddening was the consistency of the rejections. Every time Tanner's team submitted new documentation, new technical comparisons, new expert opinions, the response was essentially identical — a form letter citing the 1979 patent, with minor variations in wording. It began to feel less like a considered legal judgment and more like an automated response.

Tanner switched attorneys twice. Both new attorneys reached the same conclusion: the cited prior art did not actually cover Tanner's invention. Both attorneys filed detailed briefs to that effect. Both received the same rejection.

By 1992, the total rejection count had reached 47. Tanner had spent years of his life and most of his savings on an application that kept running into the same wall. He had been told, in effect, that his invention already existed — but every time he asked to see the proof, he was shown a document that proved nothing of the kind.

The Clerk Who Asked a Different Question

The resolution came from an unexpected direction. In 1993, a USPTO administrative clerk processing a batch of old files noticed something anomalous while conducting a routine internal audit. The 1979 patent that had been cited in Tanner's rejections — all 47 of them — appeared twice in the agency's internal filing system under two different reference numbers.

One entry was the original 1979 patent, correctly filed and accurately described. The other entry was a duplicate record created during a database migration in the mid-1980s, when the USPTO was digitizing its paper archives. During that migration, a data entry error had caused the 1979 patent's technical descriptor field to be overwritten with a description pulled from an adjacent file — which happened to describe a bracket assembly in terms that sounded, to a keyword-matching search system, very similar to Tanner's invention.

In other words, the USPTO's own internal records had been corrupted. The system was rejecting Tanner's application not because his invention already existed, but because a digital ghost of a misattributed description had been flagged as conflicting prior art every single time an examiner ran a search.

No human examiner had ever actually read both patents side by side and made a considered judgment. The automated search system kept surfacing the corrupted record, and examiners — operating under the reasonable assumption that the database was accurate — kept citing it.

What It Cost and What It Meant

Once the error was identified, the USPTO moved to correct it. The corrupted duplicate record was removed. Tanner's application was re-examined under clean search conditions and approved within four months. His patent was granted in 1994 — seven years after he first applied.

The financial damage was not fully recoverable. The USPTO acknowledged the error and expedited the final processing at no additional charge, which was something. It was not $40,000. Tanner explored legal options for recovering his costs but found that suing the federal government for a clerical error that wasn't technically classified as misconduct was its own multi-year undertaking. He eventually let it go.

The bracket assembly patent itself generated modest licensing revenue over its seventeen-year term. It was not enough to offset what the process had cost him, but it was something.

The Larger Problem

What Tanner's case exposed — and what the USPTO quietly acknowledged in an internal review conducted after the error was discovered — was a systemic vulnerability in the transition from paper to digital record-keeping. The database migration that created the corrupted record was not an isolated event. It was one of thousands conducted over several years, and the quality control procedures in place at the time were not designed to catch the specific type of overwrite error that had tripped up Tanner's application.

How many other inventors ran into similar walls and simply gave up? Nobody checked. The USPTO's internal review focused on identifying and correcting corrupted records going forward, not on auditing past rejections to see how many might have been affected.

Somewhere in the annals of American patent history, there are probably other Gerald Tanners — people who invented something real, got told it already existed, and eventually concluded that the system knew something they didn't. The system, in at least some of those cases, was just looking at its own reflection and calling it prior art.