Most disputed sinkhole claims do not begin with a hole. They begin with a cracked wall, a door that no longer latches, a floor that reads out of level, and an owner told two different things by two different inspectors. The engineering question is a differential diagnosis: a structure is deforming, several unrelated mechanisms deform structures in similar ways, and only one of them involves a subsurface void. Working that diagnosis in order separates a defensible opinion from an educated guess.
Cracks are the least reliable place to start
Crack patterns carry real information, but they are the end product of a deformation filtered through the stiffness, materials, age and workmanship of whatever was built above the ground. Two structures over identical ground movement crack differently.
Cracks are also the evidence most likely to have been altered before anyone qualified arrives — patched, painted, or widened while proving they exist. They belong in the record, mapped and measured, but as corroboration rather than primary data.
Start with the deformation shape
A relative elevation survey across the full floor area, at spacing close enough to resolve the shape rather than only the extremes, is the primary measurement. It converts anecdotes into a surface: where the structure is high, where it is low, how steeply it changes between, and whether the pattern is a bowl, a tilt or a dome.
That shape is diagnostic in a way an individual crack is not. It is also reproducible, which is why elevation data tends to anchor an investigation while crack interpretation tends to be argued about.
What void-related deformation tends to look like
Ground loss into a subsurface void is localized. The deformation it produces is typically a discrete depression, roughly circular in plan, with a steeper gradient than the structure's dimensions or loading would otherwise explain, and with a low point that need not correspond to where the building is heaviest.
The pattern frequently ignores the structure's own logic, cutting across a bearing wall or dropping one interior area while the perimeter stays near level. A deformation whose geometry is unrelated to how load is distributed indicates that something in the ground is governing.
Consolidation settlement follows the load
Settlement in compressible soil is driven by stress increase, so it correlates with the load: greatest under the heaviest elements, tapering outward, shaped by the thickness of the compressible layer. It usually produces broad, smooth dishing rather than a localized cone, and decelerates as consolidation behavior predicts.
Uncontrolled fill is the exception that muddies this, because fill boundaries are often abrupt. Settlement of fill can produce a sharp differential at the edge of the filled area, which is why establishing fill limits matters before that differential is attributed to a void.
Expansive soil moves in both directions
Shrink-swell clay responds to moisture rather than load. It heaves when wetted and settles when dried, and movement concentrates where moisture change is greatest — the perimeter, near downspouts and irrigation, near large trees. Its defining characteristic is reversibility: the same crack opens in one season and closes in another.
That signature can only be captured by measuring more than once. Atterberg limits and moisture profiles establish whether the soil is capable of the behavior; repeat surveys establish whether it is exhibiting it. Center-high doming points toward perimeter drying, not ground loss.
Not all building movement comes from the ground
A meaningful share of reported distress is not geotechnical at all. Framing lumber shrinks as it dries, long-span members deflect under sustained load, masonry and concrete shrink and creep, thermal cycling moves materials against one another, and omitted or misplaced control joints concentrate strain where cracks then appear.
Construction defects produce their own patterns — an undersized header, a missing lintel, a slab poured on unprepared subgrade — with signatures that stop at the element responsible rather than continuing into the ground. Ruling them out is unglamorous, and much of the analytical work sits there.
Time is part of the evidence
A single survey establishes a condition. Two surveys separated by a meaningful interval establish a behavior, and behavior is what most often discriminates between mechanisms: active progression, seasonal reversal, or a stable deformation that finished moving years ago.
The historical record extends the timeline backward. Sequential aerial imagery, elevation datasets, permit and repair history, and prior inspection reports often show when a depression first appeared, which bears on whether the movement predates the claimed event.
Tie the surface pattern to the subsurface
The diagnosis is not complete until the surface deformation and the subsurface investigation are correlated at the same locations. Borings placed wherever the rig could most easily reach answer a different question than borings placed at the low point of the measured depression and at a comparison location outside it.
A defined depression whose low point overlies a raveled, low-resistance soil column, which in turn overlies a solution-enlarged opening in rock, is a coherent chain of evidence. A depression with competent soil and intact rock beneath it is a coherent chain pointing somewhere else.
Where these opinions get challenged
Predictably, along a short list: that the elevation survey was too sparse to resolve the shape claimed, that no baseline exists so progression is assumed rather than measured, that expansive-soil behavior was excluded without seasonal data, that construction defects were never examined, and that the borings were not placed where the deformation is.
Work that states the measured deformation, the alternatives considered, the evidence eliminating each, and how sensitive the conclusion is to missing data withstands that scrutiny. Where both sides retain engineers, documenting the alternatives is the substance of the opinion.
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.