A derailment destroys the evidence of its own beginning as it unfolds. Cars behind the first wheel to leave the rail tear up the track structure an investigator most wants to read, and the railroad's priority of reopening the line finishes the job within hours. What survives is a set of records never designed to be read together: marks on steel and timber, geometry measurements, an event recorder, signal and train-control logs. Their value lies less in any one than in whether they agree.
Finding where the train left the rail
Reconstruction starts by locating the point of derailment, where the first wheel lost the rail. It is found physically rather than deduced: working back from where the cars came to rest, the investigator follows the marks along the track until they stop.
The distinction matters because the two zones read differently. Damage downstream describes the pileup and says little about why it began. Only the track approaching that point carries evidence of the mechanism.
What wheels write on rail and ties
Steel wheels leave a legible record on steel rail. A flange climbing the rail head marks the gauge face and crown differently from a wheel dropping inside a rail that rolled over or fractured beneath it. Off the rail, wheels chop ties and shear spikes, and the spacing of those marks records how the truck was moving.
That evidence is destroyed not by weather but by the repair — ties replaced, rail cut out, the section rebuilt, often before anyone outside the railroad has seen it.
Track geometry as it was, not as it is
Gauge, cross-level, alignment and surface are governed by the FRA track safety standards at 49 CFR Part 213, whose tolerances tighten as the class of track, and the speed it permits, rises. The measurements that matter existed before the train passed and cannot be taken afterward.
Geometry can still be measured on undisturbed track approaching and beyond the site, since a condition rarely begins and ends at one tie, and the railroad's own geometry car runs, rail flaw detection passes and documented Part 213 inspections describe the segment over time.
Rail and wheel as components
Where a rail is broken, the question is whether it broke before the derailment or because of it, and the fracture face generally answers it. A defect grown by fatigue shows an origin and progression markings a fresh overload break does not.
Wheels and bearings hold their own history: tread and flange condition compare against the condemning limits applied in interchange under the AAR rules and the freight car safety standards at 49 CFR Part 215, and a bearing that overheated in service looks nothing like one damaged in the pileup.
What the event recorder bounds
Event recorders are required under the locomotive safety standards at 49 CFR Part 229, on a crashworthy module. They capture the parameters that bound the sequence: speed, throttle and dynamic brake position, brake pipe pressure and application, and horn and bell operation.
The recorder gives a timeline of what the train did. It does not say why, and describes nothing about the track. Read alone it invites the assumption that whatever the crew did last caused what followed, when it may show only a reaction to something already underway.
Signal, dispatch and train-control records
Where authority for the movement is in question, the record sits off the train. Signal and train control fall under 49 CFR Part 236, whose Subpart I governs positive train control. Those systems produce wayside and interlocking event records, dispatcher authority records, and PTC onboard and back-office data showing what limits the system held and whether it warned or enforced.
These logs have retention limits and different custodians, sometimes different companies where trackage rights apply. Identifying the custodian early is often more urgent than analyzing the data.
The train's history before the site
A train arrives at a derailment already carrying a data trail. Wayside detectors along the route read bearing temperature, dragging equipment and wheel impact load, and their outputs exist for every train that passed, not only this one.
That supplies what a site examination cannot: a baseline showing whether a component was degrading over preceding trips, and an independent fix on timing upstream. Consist and car repair records complete the picture.
Making the records cross-check each other
The discipline lies in the comparison, not any single record. Marks on the track give distance precisely and time not at all. The recorder gives time and speed, from which distance is derived. Signal and detector logs give fixed points with their own clocks.
Aligning them requires an explicit time base, because onboard, wayside and dispatch clocks are not necessarily synchronized. Where two independent records agree on position and speed the conclusion is durable; where they disagree, the disagreement is the finding worth pursuing.
Federal control and what counsel can actually do
Access is constrained before it is technical. Where the NTSB launches, it controls the scene and the evidence, and its investigative regulations govern who participates: party status goes to organizations whose expertise the Board needs, on conditions, and is not available to litigants or their retained experts as such. The FRA investigates under its own rules regardless.
Independent examination may therefore have to wait, so preservation demands should go out immediately and be specific as to segment limits, component identities, log types and retention windows. The Board's report, useful for orientation, is barred by statute from use as evidence of fault in a civil damages action.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.