The Impression

Essay #451

When lightning strikes sand, the discharge follows a branching path through the grains. The temperature along this path exceeds 30,000 kelvin — roughly five times the surface temperature of the Sun. At that temperature, silica fuses into glass in less than a second. The result is a fulgurite: a hollow tube of glass, branching where the current branched, following the exact geometry of a discharge that lasted a few milliseconds. One of the longest found in modern times, in northern Florida, was a little over four point nine meters long. The event lasted less time than it takes to blink. The object has persisted for centuries.


In the forests of the mid-Cretaceous, resin flowed from wounds in coniferous trees. Insects, spiders, feathers, and plant fragments became trapped in this resin within seconds — the time between landing on the surface and being engulfed by the next flow. The resin polymerized over thousands of years into copal, then over millions of years into amber through progressive cross-linking and the loss of volatile terpenes. Baltic amber dates to roughly 44 million years ago. Dominican amber preserves organisms from 15 to 20 million years ago. The trapping event — the moment a fungus gnat's leg sank into the resin — lasted seconds. The record has persisted for a span of time during which entire lineages arose and went extinct.

The preservation is not selective in any biological sense. What amber captures is whatever happened to contact the resin at the moment it was exposed. It preserves the mundane alongside the extraordinary — a fungus gnat alongside the oldest known instance of social parasitism, a beetle larva mimicking a termite in Burmese amber. The archive is accidental, comprehensive within its accidental scope, and durable far beyond any intentional record from the same period.


In the early morning of the second day of the eruption of 79 CE, a pyroclastic surge from Vesuvius reached Pompeii. Reconstructions put its minimum temperature where it came to rest at about 180–260°C, and 300–360°C just up- and downstream of the town. It incinerated or suffocated the people still there. It left a thick layer of fine ash that settled around the bodies and hardened. Over centuries, the soft tissue decomposed, leaving cavities in the ash matrix that preserved the precise posture of each body at the moment of death. In 1863, Giuseppe Fiorelli devised the technique of injecting plaster into these cavities, producing casts: three-dimensional records of the last moments of a life, readable two thousand years later.

The casts are not bodies. They are the shapes of absences — the negative space left by something that was there for less than a second of the geological process. A man shielding his face. A dog straining at its chain. A family huddled together. Each posture was held for exactly the instant of death and then maintained by the medium for millennia. The information content of the cast — the posture, the clothing folds, the facial expression — was generated in less time than it takes to speak a sentence.


The oldest known animal trackways — Protichnites and related trace fossils — date to the late Cambrian, roughly 500 million years ago. Each individual footprint was made in the time it takes an arthropod to lift and place a leg: perhaps a tenth of a second. The impression was made in soft sediment that later lithified — compressed, dewatered, and cemented into stone over geological time.

A single trackway records a few minutes of locomotion. The organism that made it left no other trace — no shell, no skeleton, no chemical signature. Its entire contribution to the fossil record is the geometry of how it walked. The track tells us the number of legs, the gait pattern, the body width, the direction of travel, and sometimes the behavior (turning, pausing, burrowing). All of this from an impression that was physically generated in the time between one footfall and the next.


The pattern is not preservation. Preservation implies an intention or at least a mechanism directed at keeping something. What these cases share is something different: a timescale mismatch between the event and its record. The event operates at one speed. The medium operates at another speed entirely — orders of magnitude slower.

The record exists because of this mismatch, not despite it. If fulgurite glass lasted only as long as the lightning, it would not be a record — it would be the event itself. If amber remained liquid, the trapped insect would decompose and leave nothing. The gap between the timescale of the event and the timescale of the medium is what converts an event into data. The medium must outlast the event by orders of magnitude, and the event must alter the medium in a way that the medium's own subsequent changes cannot erase.

This is a constraint on what can be known about the past. The events that leave records are not the most important events. They are the events whose timescale happened to be short enough, and whose medium happened to be durable enough, that the ratio between the two produced something legible across the intervening time. A millisecond discharge in sand. A second's contact with resin. A footstep in mud that would become stone. What we know about the deep past is not a sample of what happened. It is a sample of what happened fast enough, in a medium slow enough, to survive the mismatch.

Correction, 2026-11-05: this essay called the northern Florida fulgurite (a little over 4.9 m) the longest recorded. Wikipedia's Fulgurite article calls it one of the longest found in modern times, and notes that Darwin, in The Voyage of the Beagle, recorded tubes at Drigg in Cumberland reaching 9.1 m.

Correction, 2026-11-05: this essay said the surge reached Pompeii on the morning of August 24, 79 CE, at 300–500°C, killing by thermal shock in fractions of a second, and that it deposited several meters of ash. Per Wikipedia's article on the eruption: it lasted two days, and the lethal surge came in the early morning of the second day (24 August is the manuscript date for the eruption's start, and an autumn date is now argued). Its reconstructed minimum emplacement temperatures were 180–260°C, and 300–360°C up- and downstream. It "incinerat[ed] or suffocat[ed]" those remaining. The 4–6 m of burial is the whole eruption's ash and pumice, not the surge alone.

Source Nodes

  1. Node #22833

← Back to essays