A cancelled show reads like a rest day. It isn’t. A drum corps season is a machine for turning repetitions into points, and the exchange rate is brutal in the last two weeks — ask anyone who has watched a corps find two points in four days in early August. Take a night out of that machine and nothing bad shows up anywhere a fan would look. The corps doesn’t drop. The gap doesn’t widen. The standings simply pause, holding a number that gets one day less true every morning.

That’s the tax. It is paid in full, it is paid by whoever happened to be on that lineup, and it is invisible.

Nine points of completion in five years is not a run of bad luck. It is the shape of a shorter, denser season colliding with a summer that keeps producing the kind of afternoon that empties a stadium: 96.0% of the schedule survived 2022, 87.0% survived 2026. And the same compression that makes the calendar tighter makes each loss cost more, because there is no longer a spare Tuesday to move anything to. In a thirty-show July you absorb a rainout. In this one you eat it.

Straight about the table: three of the five seasons are unfilled above. Measuring completion needs each year’s published schedule — the record of what was supposed to happen — and this desk holds that only for 2026. The score archive can’t stand in for it, because a cancelled show never generates a scores page and so never enters the archive at all. The endpoints are as briefed; the middle stays blank until it can be sourced rather than guessed.

The last two came back to back inside the final fortnight — Mason, OH on the 27th, Downingtown, PA on the 2th. That timing is the whole argument. A show lost on July 3 costs a corps a night they will have four more chances to replace. A show lost eleven days out costs them a night that does not exist anywhere else on the schedule, in the exact window where the sheets are tightening and every panel is worth more than the last.

Read the right-hand column twice. Every corps in that room goes -5 to -4 days between scored performances in the last fortnight of the season, and every one of them is now carrying a number into August that was written in July. The two at the top of the table are the ones that matter to the race: they came into Annapolis holding some of the best marks in the country, and they will still be holding exactly those marks when the rest of the field has been re-scored twice.

This is where the invisible tax becomes a visible one. The leaderboard on this site, the head-to-head gaps, the Monte Carlo that runs every night and prices Indianapolis — all of it reads the most recent score a corps has. It has no column for “and that score is six days old because it rained.” A corps that was climbing gets frozen at the bottom of its climb. A corps that had just dipped gets frozen at the bottom of its dip. Neither of them did anything to earn the freeze.

That 87.5% is a floor, not the league’s figure, and it is lower than the briefed 87.0% for an unglamorous reason: a scheduled show that never produced a recap on our books isn’t necessarily a show that was cancelled. It might just be one we don’t scrape. We publish the number anyway, with the caveat attached, because a metric you only quote when it flatters you isn’t a metric.

Here is the falsifiable part, and it resolves in Indianapolis. If the weather tax is real — if a lost night in the last fortnight genuinely costs points rather than just costing a Tuesday — then the corps stranded on July 28 should underperform their projections at Finals relative to the corps that competed straight through. Not dramatically. Two, three tenths, the size of a caption’s worth of cleaning that never got a public rehearsal. If they hit their projections dead on, this piece was an interesting theory about nothing, and the desk will say so.

Either way, somebody should be counting. Eight shows is not a rounding error, the number has been going the wrong way for five seasons, and right now the only place the cost appears is in a column nobody publishes.