Japan’s Fire and Disaster Management Agency has verified three deaths from Tuesday’s magnitude-7.1 earthquake beneath Kumamoto Prefecture, and reported in the same document that another 13 are being investigated. The seventeenth bulletin on the disaster, issued at 16:30 on Wednesday from the agency’s emergency headquarters in Tokyo, names the confirmed dead by municipality: two in Kashima Town and one in Kosa Town, both a short distance from the town of Uki, where the shaking registered shindo 7, the top of Japan’s intensity scale.
The 13 appear a page later, in a section of the bulletin reserved for what individual fire departments are reporting but the agency has not yet checked. The heading reads “in addition to the above”, and the entry states that Kumamoto Prefecture is reporting a large number of casualties including 13 deaths whose degree of injury and whose connection to the disaster are both still under investigation. It is not a revised toll and the agency does not present it as one. It is a note that the verified figure of three is a floor, and that the arithmetic ceiling for now is 16.
That distinction is the reason the bulletin is worth reading rather than summarising. Japan counts disaster deaths in stages: an immediate verified count, a category for deaths whose cause is disputed or unclear, and — separately, often months later — a formal determination of “disaster-related deaths” for people who die of causes triggered by the event rather than by the shaking itself. The 2016 Kumamoto earthquakes killed 50 people directly; the official toll now exceeds 270, and most of the difference was added by that third process long after the news coverage ended.
The verified injury figures in bulletin 17 are national rather than prefectural, which is easy to misread. Across four prefectures there were eight serious and 15 slight injuries, 23 in total. Kumamoto accounts for all eight of the serious injuries and 12 of the slight ones; the remaining three are single slight injuries in Fukuoka, Oita and Kagoshima. Nine fires broke out and all were extinguished, and fire crews recorded 55 rescue operations.
The evacuation is the largest number in the document. Orders covering 123,262 households and 258,206 people were issued across 43 municipalities, with emergency safety instructions — a higher-urgency category — covering a further 46,610 households and 92,743 people in two more. A national emergency fire response force of 1,410 personnel was in the prefecture as of Wednesday afternoon, drawn from units outside Kumamoto. Rescue crews were still working three sites: a collapsed factory chimney in Yatsushiro, the collapsed second floor of a commercial building in Kashima, and two buckled houses in Hikawa.
One half of the bulletin’s damage table is simply empty, and that is informative too. Alongside the casualty columns sit six for housing — destroyed, half-destroyed, flooded above and below floor level, partially damaged, total — and for Kumamoto Prefecture every one of them is blank more than a day after a shindo-7 earthquake. The only dwelling damage recorded anywhere in Japan is four partially damaged homes in Fukuoka, a prefecture that logged a single minor injury. Damage assessment has not begun in the area that was actually shaken; the surveyors are still doing rescue.
One apparent contradiction in the record resolves on inspection. Japan’s Meteorological Agency has catalogued 233 earthquakes across Kumamoto Prefecture’s three seismic regions in the 35 hours after the 16:27 mainshock — a count that was already 237 a few hours later, because the list updates continuously. Query the US Geological Survey’s catalogue over the same window and the same area and it returns ten. The two agencies are not disagreeing about what happened; they are running instruments at different densities. Above magnitude 4.5, where both networks detect everything, JMA lists six events and USGS seven, and five of those are the same earthquakes. Everything below that threshold is picked up by Japan’s dense domestic network and by almost nothing else.
Where the two catalogues genuinely differ is on size, and by a consistent margin: JMA put the mainshock at M7.1 against the USGS figure of 6.8, and the largest aftershock at M5.8 against 5.4. Those are different magnitude scales rather than competing measurements of the same one, a distinction this site set out when the earthquake struck. The gap has held at roughly three to five tenths of a unit across every event large enough for both to have measured it.
Two aftershocks arrived after bulletin 17 went to press, the larger of them a magnitude 5.8 at 22:19 on Wednesday night. Neither had been reflected in the casualty figures at the time of writing, and the agency states on the front of every bulletin that the numbers are preliminary and subject to change.