The last thing this site published about the earthquakes that struck Venezuela on 24 June was that the death toll had passed 1,700. That was on 30 June, and we did not return to the story. The United States Geological Survey's impact record for the event, last revised on 28 July, puts it at at least 5,500 people killed, 16,700 injured and at least 29,995 missing.

The gap is not a dispute between sources. It is the difference between a count taken six days after a disaster and one taken five weeks later, and it is the ordinary arithmetic of an event whose victims were still being dug out. We are publishing the later figure because we published the earlier one and left it standing.

The two earthquakes arrived 32 seconds apart. The USGS catalogue records a magnitude 7.2 at 22:04:31 UTC, 20 kilometres east of San Felipe, and a magnitude 7.5 at 22:05:03, 17 kilometres west of Catia La Mar on the coast road between Caracas and its airport. Both were ten kilometres deep, which is shallow enough to put the full force of the rupture into the ground surface directly above it.

The agency's own automated loss model, PAGER, went to a red alert for both fatalities and economic losses — its highest level, and one it describes as reserved for events that have historically required a national or international response. Its exposure tables put roughly 1.25 million people in the zone of severe shaking and a further 117,000 in the zone above that. Its economic estimate is expressed as a share of national output rather than a dollar figure: between 5 and 20 per cent of Venezuela's GDP.

The damage record is concentrated in a narrow coastal strip. USGS logs at least 28,300 homes destroyed across the Caracas–La Guaira area, and in La Guaira itself — the port town directly above the larger rupture — at least 80 per cent of buildings destroyed. Elsewhere the toll is legible state by state: 19 killed and 568 homes destroyed in Carabobo, 18 killed in Aragua, 12 killed in Falcón, 25 homes and two buildings destroyed in Yaracuy.

One entry in the file has nothing to do with buildings. Alongside the tsunami readings — four centimetres at Isla de Mona and Yabucoa in Puerto Rico, two at Fort-de-France in Martinique — the record notes coastal uplift of 6.1 metres in the Galfa Point area of Trinidad. That is the seabed and shoreline permanently lifted by about the height of a two-storey building, some 500 kilometres from the epicentre.

The missing are the figure to watch. Thirty thousand people unaccounted for against five and a half thousand confirmed dead is a ratio that usually resolves in one of two directions over the following months: downward, as the displaced are traced and duplicate reports are reconciled, or upward, as rubble is cleared. Which way it moves is not yet in the record, and this site will not guess at it.

What the record does establish is that the number we left on the site in June was wrong by a factor of three, and stayed there for five weeks.