The trade court is having its second-heaviest year since 1999, not its heaviest. And our own July count of the 2020 wave was 220 cases short

The US Court of International Trade has docketed 4,611 cases since the fiscal year opened on 1 October 2025, with five weeks of it still to run. Counted the way the Administrative Office of the US Courts counts — its published series runs a little under a raw docket count, by nought to four per cent in each of the last twelve fiscal years — that is roughly 4,530 cases, and it makes fiscal 2026 the second-heaviest year in the thirty the Administrative Office's series covers. It is not the heaviest. Table G-1 for the twelve months ending 30 September 1999 prints 6,433 cases filed, against 3,575 the year before, in the wave that followed the courts striking down the Harbor Maintenance Tax as applied to exports. Fiscal 2020, the year usually named as the modern precedent, took 3,670.

That matters here because this publication said otherwise. Our 31 July story was headlined “Tariff refunds have reached $119bn, and the trade court is taking cases at a pace it has seen once before”, and the once was 2020. The comparison was drawn from docket data reaching back to 2013, and within that window it was right; the sentence it produced was not, because the court has seen a heavier year, and by a wide margin. The same story also carried an arithmetic error of its own. It said 3,102 of 2020's cases were filed in five days in September, then printed the five days: 220 on the 17th, 1,281 on the 18th, 665 on the 19th, 434 on the 20th and 722 on the 21st. Those add to 3,322, which is also what the docket says. The 3,102 was wrong on the page's own evidence, and it was repeated in the story's key points.

The rest of that piece stands. Filings really have run above 114 a month every month since December 2025, the March peak of 1,277 is real, and the point the story was making — that 2020's spike was a single filing deadline while this is a sustained eight-month run — survives the correction and is sharpened by it, because fiscal 1999 was a deadline-driven wave too. One caution for anyone rebuilding these numbers: the docket index keeps backfilling. The figure was 4,609 when this count was first taken and 4,611 a few hours later, and a comparison we had considered running — that 2026 beat the same January-to-August window of the previous sixteen years combined — has decayed from a margin of 161 cases on 10 August to nine today, and is false at seventeen years. It was the largest span that worked, which is reason enough not to print it.