Observation log, September 2026

This is the first of these. Each month FlyGamma will publish what its feed actually recorded, counted the same way every time, so the numbers can be set against one another. On its own a single month says little. The value is in the series, and a series has to start somewhere.

What the first two days looked like

The shape of the opening is mostly the shape of a new archive learning the world rather than of anything happening in aviation. Routes went from 821 to 14,008 and aircraft from 1,325 to 21,917 in a day, because almost everything the feed described was something the archive had never seen before. That rate will fall away sharply: there is a finite set of regularly flown city pairs, and once the common ones are recorded, later passes mostly re-confirm what is already known rather than adding to it. The interesting figure in a month's time is not how fast it grew but where it settled.

Two things about these numbers are worth stating plainly. Nearly half of the 14,008 routes have been observed exactly once, which is what a two-day window looks like — a great many pairs will have been caught on a single pass and not since. And 2,918 airports is a larger number than it sounds, because it counts any airport seen at either end of any observed flight, including a great many that were seen once.

The busiest things seen

The most-seen route pair was Kuala Lumpur to Singapore, with 135 sightings. That is a good illustration of why these are sightings rather than services: the flight takes roughly forty-five minutes, the sampler runs at least forty-six minutes apart, and so an unknown number of those rotations were never caught at all. The pair ranks first despite the sampling working against it.

After it came New York–Los Angeles (106), Jakarta–Singapore (105), Nantucket–Boston (92), Jeju–Seoul Gimpo (83) and Jeddah–Riyadh (72). Four of the six are short-haul pairs in dense regional markets, which is where the sampling gap bites hardest — so if anything those are the most understated numbers on the list.

By airline the order was Delta (3,736 sightings), American (3,292), United (2,740), Southwest (2,291), Ryanair (2,285) and British Airways (1,138). The American weighting is real but overstated: the wide feed's coverage is denser over North America and Europe than elsewhere, so this is a fact about the supplier's reach as much as about world aviation.

By airport, Atlanta led, followed by London Heathrow, New York JFK, Chicago O'Hare, Singapore Changi, Dallas–Fort Worth, Denver and Frankfurt.

Changes to the method this month

Three, all made in the first days and all disclosed here because they affect how later months compare with this one.

The archive began recording when it first met each airline, route, airport and aircraft. Entries recorded before that change carry no first-seen date, and deliberately have not been given one — the archive does not know when it met them, and inventing a date would be the sort of thing this site exists not to do.

The ceilings on the route and aircraft tables were raised, from 30,000 and 50,000 to 60,000 and 120,000. At the rate above, the old ones were days away.

The rule for what gets discarded when a ceiling is reached was changed. It kept whatever had been seen most recently, which would have meant discarding a pair seen five hundred times over a month in order to keep one seen once, five minutes ago. It now keeps what has been seen most often.