Why buildings open up: how the gaps pests use are actually made

The building envelope · reviewed August 2026

Why is an opening better understood as a process than as a hole?

Because almost nothing a rodent uses to get into a building was built as an opening. It was manufactured, slowly, by the structure doing ordinary things — taking up water and giving it back, warming and cooling, settling, and being repaired by people who were solving a different problem.

That distinction sounds academic until you watch what it does to a survey. Somebody who believes a gap is a hole looks for holes, finds four, closes them, and is surprised the following autumn. Somebody who believes a gap has a history looks instead for the places where the building is producing gaps: the joint that moves seasonally, the elevation that stays wet, the run of mortar that a previous owner had repointed with the wrong material, the run of trim that was pinned rather than fastened. Those places will produce a second opening whether or not you closed the first one.

Three engines account for nearly all of it, and they compound. Movement opens joints. Water softens what the joint is made of. Repair, done in a hurry with an adhesive product, buys a season and then hands the animal a better edge than it had before. Everything below is an unpacking of those three, and the reason it belongs on a pest control site rather than a builder’s is that the same three engines decide where the animals will be next year.

The practical yield is a different question. Not where is the hole, but what in this wall is making holes, and what would a closure here have to survive. Answer those and the work stops being a patching exercise. Answer neither and the building is on a maintenance schedule that never converges.

What is actually moving in a building, and by how much?

More than most people assume, and mostly in one direction.

Framing timber is hygroscopic. It reaches an equilibrium moisture content with the air around it, and that equilibrium tracks the seasons. The Forest Products Laboratory’s tables put wood at roughly 9.2 to 9.5 percent moisture content at 68 to 70 degrees Fahrenheit and 50 percent relative humidity, rising to about 11 percent as humidity climbs to 60. The laboratory states the mechanism plainly: wood in service is exposed to seasonal and daily changes in the relative humidity and temperature of the surrounding air, and those changes induce changes in the wood’s moisture content.

What matters for our purposes is the direction of the resulting movement. Wood shrinks and swells most in the direction of the annual growth rings, about half as much across the rings, and only slightly along the grain. So a stud does not get meaningfully longer or shorter over a year, but a sill plate, a band joist and a piece of exterior trim all get measurably thicker and thinner. Every joint where a wood member meets something that is not wood — a foundation wall, a steel lintel, an aluminum flashing, a masonry chimney — is therefore a joint that opens and closes on an annual cycle while the members either side of it stay where they are.

Now add the fasteners. A nail in a member that swells and shrinks around it withdraws fractionally each cycle; anyone who has looked at forty-year-old cedar siding has seen the row of proud nail heads that results. Add thermal movement, which runs at different rates in metal, masonry and timber, so a metal drip edge lapped over a masonry course is in relative motion with it for the whole of its service life. Add differential settlement, which the National Park Service names among the root causes to correct before masonry repair is even attempted.

None of this is failure. It is the ordinary behavior of an assembly made from materials that respond differently to the same weather. But it means that a joint sealed with something sticky is being asked to stretch and relax several thousand times, and that a gap at a sill plate is better read as a seasonal dimension than as a defect somebody caused.

Why does the mortar open before the masonry does?

Because it was designed to. This is the single most useful thing a homeowner with an older foundation can know, and almost nobody is told it.

The Park Service sets out the principle in its guidance on repointing: historically, mortar acted as a bedding material — not unlike an expansion joint — rather than as a glue holding the units together, and the stresses in a wall are meant to be relieved by the mortar rather than by the masonry units themselves. The joint is the sacrificial element. It is supposed to take the movement, take the water, and give way first, because a joint is repairable and a spalled brick face is not.

On the North Shore that plays out over decades of freeze-thaw cycling on fieldstone and rubble foundations whose joints were never continuous to begin with. Water gets into a joint, freezes, expands, and takes a little of the joint with it each time. What you end up with is not a neat hole but a tapering void that narrows as it goes and then opens out again into the wall — which is exactly why a surface patch on a rubble foundation is theater. The patch spans a void it never filled, and the first animal to test it finds a soft margin at the perimeter.

The failure mode that follows is worth naming because it is common and it makes things worse. Somebody repoints an old soft-brick or rubble wall with a hard modern cement mortar. The Park Service records the consequence directly: moisture trapped in the wall was then unable to evaporate through the mortar that was intended to be sacrificial and to protect the brick, and the water pressure eventually popped the surface off the bricks. The building now has spalled faces — rough, friable, edged surfaces — at exactly the working height of a rat, and the joint that used to relieve the movement no longer can.

So on an older masonry building, the question is not only where the joints are open. It is whether the last repointing campaign made the wall more able to shed water or less.

What actually makes an opening gnawable?

An edge. That is the whole of it, and it reframes the dimensions everyone quotes.

The published figures are worth having: the University of Nebraska exclusion handbook states that rats crawl through or under any opening higher or wider than half an inch, and that mice enter openings larger than a quarter of an inch. Those numbers are correct and they are the reason a door sweep matters — New York City wrote the same fraction into law, requiring an owner correcting a pest condition under Local Law 55 of 2018 to fit sweeps that close the gap beneath a door to no more than a quarter of an inch, alongside sealing holes and cracks around pipes and conduits.

But the same handbook is explicit about the limit of those numbers. Rodents can quickly gnaw into most materials when they are given a rough surface or an edge to bite into, and the list of materials they get through includes lead and aluminum sheeting, window screens, wood, rubber, vinyl, fiberglass, plastic, and low-quality concrete or concrete block. Read those two statements together and the dimension becomes a starting condition rather than a finishing one. A quarter-inch gap is not a barrier that holds at a quarter of an inch. It is an invitation with a handhold.

Which turns the specification question inside out. The useful property of a closure material is not only its hardness but whether it offers anything to grip: a lip, a soft margin, a raised bead, a cut end, a corner that stands proud. A hard, flat, continuous surface with its edges buried or clamped gives an animal nothing to start on. A material that is soft all the way through gives it purchase everywhere.

Which is the honest reason this firm will not put expanding foam into an assembly in any role. Foam is edge and purchase throughout, it goes in fast on the afternoon when the correct detail is slow, and — worse than either — it reads as finished work to the next person who opens that wall. It is the trade’s most common closure and its most common failure. We do not use it as a barrier, as backing behind one, or as a seal over one.

Why is height so rarely a defense?

Because the animals are better at architecture than the survey usually assumes, and the figures are not close.

The Nebraska handbook credits rats with a vertical jump of about 36 inches from a flat surface and a horizontal jump of as much as 48 inches. It puts their reach at as high or as wide as 13 inches, and notes that they climb the outside of vertical pipes and conduits up to three inches in diameter and can climb inside pipes between roughly one and a half and four inches. Mice jump as high as 18 inches from a floor onto an elevated surface, and can travel considerable distances crawling upside down along screen wire.

Stack those capabilities against a real elevation. A downspout is a climbable three-inch vertical. A fence returning to a wall is a launch platform. A window air-conditioner is a shelf. A row of clapboards is a ladder to anything with a reachable edge, and an unclosed soffit return sits at the top of it. The consequence is that an opening two stories up is not out of scope; it is merely inconvenient to inspect, which is a different thing and is why it is so often skipped.

It also explains a pattern that puzzles homeowners: a building whose ground floor has been thoroughly sealed and which now has activity in the attic. Nothing has changed except that the easy route closed and the animals used the one they always could have. Sealing from the ground up without ever getting above the gutter line produces exactly this, reliably enough that we treat a report of new overhead noise after ground-level work as confirmation the ground-level work was real.

The survey therefore has to be three-dimensional and it has to be done from a ladder or a lift rather than from the lawn with binoculars. Cornell’s exclusion guidance is written the same way, treating the roofline, vent screening and chimney closure as part of the same continuous barrier as the foundation.

Why does the previous repair so often become the next opening?

Because most repairs are adhesive, and adhesion is a bet on a surface nobody prepared.

Look at what is typically holding a repair together on a residential elevation. A bead of caulk laid into a joint that was damp, dusty or chalking. A patch of mortar buttered over a joint that was never raked back to sound material. A length of trim pinned at the corners over a gap that was not measured. A piece of screen stapled at its perimeter rather than fastened through a frame. Each of those is relying on a bond to an unknown substrate, in an assembly that is going to move seasonally for the rest of its life, outdoors, in weather. The failure is not a question of if.

And the way each one fails hands the animal something better than it had. Caulk hardens and releases from one face of the joint, leaving a flexible flap. Mortar buttered over a dirty joint debonds and stands proud, leaving a lip. Trim pinned at the corners bows in the middle, leaving a shadow line and a leading edge. Screening stapled at the perimeter tears at the staples, leaving wire ends. In every case the repair has converted a smooth opening into an opening with a handhold, which is why the second-generation gap at a given location is usually easier for the animal than the first was.

There is a second failure class that is more damaging and less obvious: the repair that closes something which was doing a job. A weep hole mortared shut, a crawlspace vent taped over, a soffit vent blocked with insulation, a dryer duct sealed at the wall because the flapper rattled. Each of those was a well-meant closure, and each traps water in an assembly that was designed to shed it. Give that a few seasons and the wall produces decay, and decayed wood is soft wood, which brings its own tenants and its own openings. A closure that creates a moisture problem has not saved you a repair; it has scheduled a larger one.

So the most productive twenty minutes on most surveys is spent looking at what the last person did, not at what the building did.

Which openings are supposed to be open, and what does that change?

Rather more of them than people expect, and it changes the whole approach from sealing to screening.

Brick veneer is a drained assembly: water gets past the outer wythe by design, runs down the back of it, lands on flashing and leaves through the weep holes at the bottom of the cavity. Those holes have to stay open. Crawlspace and foundation vents move air through a space that would otherwise sit at high humidity against timber. Ridge and soffit ventilation work as a pair and depend on both ends staying clear. Combustion air openings feed appliances. Dryer and bath exhausts have to discharge, and their dampers have to be free enough to close when the fan stops. Condensate lines and sump discharges have to drain.

For every one of those, the correct closure keeps the function and adds the barrier. That means mesh, and it means choosing the mesh deliberately. Cornell’s exclusion guidance describes quarter-inch and half-inch hardware cloth as the working sizes, notes that half-inch is stronger but less flexible than quarter-inch, and that quarter-inch is the more appropriate choice for smaller vertebrates such as mice and bats. It also observes that galvanised sheet metal is durable and, when attached with screws, resists removal by an animal with hands — which is the reason a soffit return closure gets fabricated metal and fasteners rather than a panel pushed back into place.

The distinction between sealing and screening is the detail most often got wrong on a residential job, and it is got wrong in both directions. A weep hole packed with mortar is a sealed drain. A crawlspace vent left with rusted-through screening is an open door. The same opening, treated by two different people, fails two different ways.

There is a corollary worth stating for anyone reading their own building: if you cannot say what an opening is for, find out before you close it. That question has saved more walls than any product on the truck.

What does a closure have to survive before it counts as permanent?

Four things, and a closure that cannot survive all four is a repair with a date on it.

It has to survive movement — the seasonal cycle described above, which means it must either span the joint with a rigid member fixed to sound material either side, or accept the movement without opening. It has to survive water, which means it cannot trap moisture in an assembly designed to drain, and it cannot itself corrode into a stain down the elevation. It has to survive teeth, which means hard, continuous and edgeless where an animal can reach it. And it has to survive the next trade — the roofer, the electrician, the cable installer, the mason — which means it has to be legible as deliberate work rather than as something to cut through.

That is the whole reasoning behind a short and slightly unfashionable materials list. Custom-fabricated 26-gauge galvanised sheet metal, cut and bent to the opening rather than trimmed from a stock piece and made to look close enough. Cement and mortar at masonry, matched to the wall rather than to whatever is strongest. Xcluder door sweeps and fill fabric at thresholds and around service penetrations, and Xcluder GEO where a barrier has to sit below grade against burrowing. Copper and stainless mesh where an annulus has to be packed and a corroding fill would stain the wall beneath it. Quarter-inch hardware cloth at any opening that still has to breathe. And mechanical fasteners throughout, because a screw into sound material is a fact and an adhesive bond is a hope.

None of that is exotic and none of it is proprietary magic. It is simply the list you arrive at once you accept that the closure has to outlive the conditions that made the opening. The structural exclusion page sets out how the work is scoped and sequenced; this piece is about why the specification looks the way it does.

Why can a building not be closed while animals are still inside it?

Because sealing a structure that still holds a population removes the exit, not the population, and everything after that gets worse.

Cornell’s guidance builds the caution into the method: monitor an entry site for at least two days before you seal it, so that you know whether it is in use. It also notes the seasonal traps on either side of the calendar — animals are inactive in winter but will emerge during a warm spell, and spring and summer work is complicated by the presence of dependent young. Close an opening on a nursing female and you have created an animal welfare problem and an odor problem in the same afternoon, and neither is solved by the closure you just installed.

There is a structural reason too, and it follows directly from the edge argument. An animal working from the inside of a wall has access to substrates that were never intended to face an attack: the back of a soffit panel, the paper face of insulation, the unfinished edge of sheathing, the soft side of a subfloor. Every one of those presents an edge. A closure engineered to defeat an animal outside the envelope is frequently trivial to defeat from within it, which is why sealing first and reducing later is not merely out of order but actively counterproductive.

So the sequence is fixed: read the evidence, bring the population down, confirm the structure is empty, then close it and verify the closure after a weather cycle. Rodent evidence is the tool for the first and third of those steps, and the rodent control page describes how it is read. The house mouse and Norway rat profiles carry the identification detail.

How should you look at your own building with this in mind?

Look for change, not for holes, and start at the last repair.

Walk the perimeter slowly at ground level and read the wall rather than scan it. Mortar of a different color is a previous campaign; note whether it stands proud of the brick and whether the faces near it are spalling. A dark tail below a joint is water leaving where it should not. Rust weeping from a fastener head means the fastener has been wet for years, and what it is fastened to has been wet with it. Paint failing in a line rather than in a patch is following a joint. A bulge in siding is something behind it that got wet and moved. None of these is a pest observation and all of them predict one.

Then look up, properly, with a ladder or from a neighbor’s vantage. Soffit returns at every corner where a roof edge dies into a wall. Gable louvers and the state of their screening. The ridge line. Fascia and drip edge. The chimney crown. Then look at the service entries — water, gas, electric, refrigerant lines, irrigation, cable — and ask of each whether the sleeve was ever packed and what packed it.

Finally, list what you found in two columns: openings that are supposed to be open, and openings that are not. The first column gets screened, the second gets closed, and nothing in either column gets a product that a rodent can bite. Everything else in this cluster is indexed on the blog, the pest library puts names to what you find, and where the problem is a building that has not been built yet, that is a design conversation and it lives with exclusion consulting rather than here.

The work behind this piece is Structural Exclusion, and everything else in this cluster is indexed on the blog hub.