How to find the source apartment in a cockroach line

Plumbing chases, risers and service cores · reviewed August 2026

Why do complaints in a building arrive in a vertical pattern?

Because the habitat is arranged vertically, and the complaint log is a distorted picture of it.

Pull three years of service records for an eighty-unit building with a German cockroach problem and the calls will not be scattered evenly. They fall in a column. The B line reports in March, reports again in June, then an apartment two floors below the first one reports in September — while the A line on the same landings, with the same age of fit-out, the same refuse room and the same corridor, has said nothing at all. Most managing agents have noticed this pattern. Very few draw the conclusion it demands.

The conclusion is that whatever produces the insects is laid out the way the plumbing is laid out. Residential buildings stack their wet rooms on purpose, because running one riser past twelve kitchens is far cheaper than running twelve risers to twelve kitchens. The kitchen in 2B therefore sits directly under the kitchen in 3B, which sits under the kitchen in 4B, and behind every one of them is a single continuous cavity carrying cold and hot supply, a waste stack, a vent stack and — in a great deal of the pre-war stock in Manhattan, Brooklyn and Queens — a heating line that keeps that cavity warm from October to April.

Nothing about a household’s habits is distributed that way. Nobody’s cleaning standards travel up a pipe. When a phenomenon follows the shape of the building’s mechanical services rather than the shape of its social arrangements, the building’s mechanical services are the better explanation, and that is as close to a controlled experiment as anyone gets in occupied housing.

There is a second observation buried in the same records, and it is the one this piece is really about. The unit that reports first is not usually the unit with the most insects. It is the unit with the most attentive occupant. A retired resident who is home all day and gets up in the night will file a complaint at a population level that a household of shift workers would not register for another eight months. So the complaint log measures vigilance and confidence — two things that vary enormously between households — and only very loosely measures where the cockroaches are.

Work from that log and you will treat the vigilant apartments repeatedly, watch them clear and then refill, and eventually conclude that cockroach work does not hold. The program was never wrong about the material. It was wrong about the address.

What is actually behind the kitchen wall?

A void, and usually a much larger and better-connected one than the resident imagines.

Open the cabinet under a kitchen sink in a pre-war building and you are looking at the back of a partition that was never intended to be a finished surface. The waste line comes through it in a hole cut oversize with a hammer and a cold chisel, because the fitter needed room to swing the pipe into place. The hot and cold supply arrive through two smaller holes, each dressed on the room side with a thin chrome escutcheon that covers the hole without sealing it. Behind that partition, the space runs sideways into the adjoining bay and downward past the floor framing into the ceiling void of the apartment below.

That is the essential point about a chase and the one that is hardest to picture from inside a finished room. It is not a duct with walls. It is the absence of construction — a continuous three-dimensional gap that exists because nobody built anything there, threaded with pipes that provide condensation on the cold lines in summer and radiant heat from the hot lines all year. The University of Kentucky’s cockroach guidance makes the practical version of the point plainly: German cockroaches travel between rooms and between adjoining apartments by way of walls, ceilings, pipes and wires.

Now add what the insect wants. Cornell IPM describes cockroaches as cryptic animals that prefer tight, dark spaces — gaps, crevices and corners. A pipe passing through a rough hole in a masonry partition produces, along its entire length, exactly the geometry the animal seeks: two surfaces meeting at a narrow angle, in the dark, above ambient temperature, within a few feet of water. There is no need to imagine that the population is doing anything clever. It is sitting in the best available real estate in the building, and the building supplied it for free.

The food supply is the least demanding part of the arrangement. Grease film behind a range, residue in a trap, the glue on stored cardboard and the crumbs in a floor joint will each sustain a colony, and the animal only has to leave the void far enough to reach one of them. A female’s productivity does the rest — the Kentucky guidance notes that up to forty nymphs emerge from one German cockroach egg case, and that a single mated female can put thousands of descendants into a building inside a year.

So the void is warm, wet, dark, tight, connected to eleven kitchens and adjacent to food. The apartment is not the habitat. The apartment is where the habitat has an opening.

Why is the complaint list a bad map of the infestation?

Because it systematically misses the units that matter most, and there is published work putting a number on how badly.

A team working in a 188-unit high-rise ran building-wide inspections with sticky monitors rather than relying on what residents said. At baseline, 49 per cent of the apartments they trapped were infested, with an average of 54 cockroaches caught per infested unit. That is roughly half a building. No management office in the country receives complaints from half its apartments.

The follow-up figure is the one worth reading twice. At the twelve-month inspection, of the residents whose units the traps showed to be infested, 88 per cent did not know they had cockroaches. Not “did not report” — did not know. The paper’s conclusion is blunt and it is the operational heart of this article: placing traps was necessary in order to detect the infestations at all.

Consider what that does to a complaint-driven program. If nearly nine in ten infested households cannot tell you they are infested, then the units generating your call volume are a small, self-selected and unrepresentative sample of the units generating your cockroaches. Treat the sample, and the population in the unsampled units keeps feeding the chase. The line refills. The board is told the problem is “coming back”, when nothing ever went away — the reservoir simply moved back across the threshold of visibility in the apartments where somebody was looking.

There is a second-order effect that makes it worse. A resident who reports is often a resident who has already been treated once or twice, so the units in the log are the units with the lowest populations and the best cooperation. The crew arrives, finds very little, and reports an improving situation. Meanwhile the unit with the fifty-cockroach trap count is three floors up behind a door nobody has been through in two years.

None of this is a criticism of residents. A moderate German cockroach population is nocturnal, spends the day pressed into cracks inside cabinetry and appliance housings, and produces no noise, no smell and no visible damage. Expecting a household to detect it reliably is expecting them to do a job that requires equipment. Which is the whole argument for the equipment.

How is a line actually mapped?

By trapping every unit in the column on the same schedule and writing the numbers down, which is far more boring and far more decisive than it sounds.

The mechanics are simple enough to describe in a paragraph. Sticky monitors go into standardized positions in every apartment in the line: in the cabinet under the kitchen sink, beside the range, beside the refrigerator, and next to the toilet. Standardized matters more than clever placement, because the whole exercise depends on the numbers from 6B being comparable to the numbers from 2B. They stay down for a defined interval, they are collected rather than left to age, and the count from each position is recorded against the unit, the date and the location.

Position turns out to matter a great deal. In the high-rise study, traps set beside the stove and the refrigerator caught significantly more cockroaches than traps set in the cabinet under the sink or in the bathroom. That is a practical warning about the most common shortcut in the trade: one trap per apartment, dropped under the sink because that is where the inspector was already kneeling. A single monitor in a below-average position turns an infested unit into a clean data point, and one false clean unit in the middle of a column is enough to send the investigation to the wrong floor.

What the completed grid gives you is not a yes-or-no answer but a surface. Counts by unit, plotted against the floor, usually show a peak and a decline in both directions from it. That shape is the finding. It says the population has a center of gravity, and it says where.

The same paper looked at the spatial statistics directly, and its conclusion is worth stating carefully: at the outset the infestations were spatially correlated across the building, and after twelve months of building-wide work only the correlation across floors remained. Read that as a description of what survives. Horizontal spread between neighbors on a landing is comparatively easy to interrupt. The vertical connection — the riser — is the one that persists, because it is a structural feature and nothing about a treatment program changes it.

The grid also does something no complaint log can. It records the units where nothing was caught, which is the only defensible basis anyone ever has for saying an apartment is clear.

What separates a source unit from a receiving one?

Trap count is the first signal, but the composition of the catch and the direction of the gradient are what actually settle it.

Numbers. A unit catching in the dozens sits at a different point in the system from a unit catching two. Two is a receiving unit: insects arriving along the chase from somewhere with a bigger population. Dozens is a resident breeding population with harborage of its own.

Life stages. A catch made up largely of small nymphs, with egg cases present in the harborage found on inspection, indicates reproduction happening in that apartment. A catch of adults and large nymphs with no oothecae anywhere is more consistent with animals that walked in. Nymphs do not travel as far or as fast as adults, so a nymph-heavy catch is a strong statement about where the insects were born.

The gradient. Plot the counts up the column and look at which way they fall. A source apartment sits at a local maximum with lower counts above and below it; a receiving apartment sits on a slope. Where the numbers rise as you go down, look at the cellar, the refuse room and any commercial tenancy at the base of the riser before assuming the answer is residential at all. In a mixed-use building the ground-floor kitchen and the residential stack above it are connected by the same penetrations, and the restaurant is frequently the bottom of the column.

Position within the unit. Where in the apartment the catch is concentrated is a directional clue. Heavy counts confined to the wall shared with the chase, falling away toward the front of the apartment, describe an insect entering at the riser. Counts spread across the kitchen, into small appliances and up into wall cabinets, describe a population that has been established long enough to expand away from its entry point.

What the resident says, weighted properly. A household reporting daytime sightings in open floor space is telling you something real: that is generally a sign harborage is saturated. It belongs in the file. It just does not outrank a trap count.

Put those together and the source unit usually identifies itself within one trapping cycle. In our experience it is very often an apartment that has never called, occupied by somebody who is either uninterested, embarrassed, elderly, working nights, or afraid that reporting will be held against them — which is why access, not technique, is the binding constraint on this kind of work.

Why does the housekeeping theory keep sending crews to the wrong apartment?

Because it is intuitive, it is socially satisfying, and it fits a real correlation backwards.

The correlation is genuine: heavily infested apartments do, on average, have more clutter, more accessible food residue and more stored cardboard than lightly infested ones. What people then infer is that the mess caused the infestation. The more defensible reading is that a building supplied the population and the conditions inside a given unit decided how large it grew there — carrying capacity, not cause. Cornell’s guidance lists both halves as jobs to be done, removing the clutter and the food that sustain pests and sealing the gaps that allow entry and harborage, and it is the second half that a building consistently skips.

Three things go wrong when the first theory takes hold in a building.

The investigation goes to the wrong address. Someone walks the line, forms a view of which household looks least organized, and that unit becomes the working hypothesis. Sometimes it is right. Often it is a unit with a modest population and a lot of visible clutter, while the actual source is a tidy apartment with a chronic leak under the sink.

Reporting stops. A resident who has been told their kitchen is the reason twelve families have cockroaches does not call the office again. In a building where 88 per cent of infested households already do not know they are infested, a policy that discourages the remaining 12 per cent from speaking up is close to the worst available option.

The structural work never gets commissioned. If the diagnosis is behavior, the remedy is a letter, and a letter costs nothing. The chase stays open, the riser keeps distributing, and next spring the building writes the letter again.

The honest position is narrower and more useful. Sanitation improves bait uptake, lowers the ceiling on population growth, and makes inspection possible. It cannot dry a chase, it cannot warm-proof a riser, and it cannot stop insects arriving through a hole in a party wall from an apartment whose occupant has done nothing wrong either. Anyone selling a building a cleaning schedule as a solution is selling the cheapest thing in the catalog.

What does closing a chase involve, floor by floor?

Sealing every penetration on every floor of the column, in materials matched to the substrate, with the sequencing that lets the treatment work first.

The order is not negotiable. Monitors first, so the map exists. Population reduction next, so that the animals sealed inside a void are not a substantial breeding colony with nowhere to go. Then the physical work, floor by floor, and then continued monitoring to confirm the void has stopped delivering.

The inventory at each floor is short and repetitive, which is what makes it auditable:

The waste-line annulus. The oversize hole around the drain where it passes into the wall or through the floor. Masonry substrate takes cement and mortar; where the pipe has to remain free to move, the gap is packed with Xcluder fill fabric and closed rather than mortared solid against the pipe.

Supply-line escutcheons. Every chrome plate on every fixture in the line, lifted, the hole behind it packed and sealed, the plate reseated. This is the single most commonly skipped item in the trade and there are typically six to ten of them per apartment.

Cabinet backs and toe kicks. The void behind the carcass and beneath the base units connects to the wall cavity in most fitted kitchens. Closing it removes harborage and cuts the route in one operation.

Conduit, cable and outlet boxes. Electrical and low-voltage penetrations pass through the same partitions and are sealed the same way.

Chase access panels and shaft openings. Where the building has a formal shaft, the openings at each floor get closed with custom-fabricated 26-gauge galvanised sheet metal, mechanically fastened, and mesh where an opening has to keep ventilating.

Foam does not appear anywhere on that list, in any role. It is the industry’s default closure and its most common failure, it is soft enough to be reopened, and once an opening is foamed it photographs as finished and is never inspected again. This firm does not use it as a barrier, as backing behind one, or as a seal over one, and the structural exclusion page sets out what goes in instead.

There is a compliance dimension in the city that boards should know about. Local Law 55 of 2018 requires an owner to take reasonable measures to keep the premises free from pests and indoor allergen hazards, and it names the physical work explicitly: eliminating points of entry and passage for pests by repairing and sealing holes, gaps and cracks in walls, ceilings, floors, molding, base boards, and around pipes and conduits, along with door sweeps closing gaps to no more than a quarter of an inch. The statute describes a sealing program. A building running bait alone is not doing what the law asks of it.

What should a board ask for, and how do you know when it is finished?

Ask for the map before the treatment plan, and judge the ending on trap counts rather than on silence.

Four things belong in the scope from the beginning. First, the unit of inspection is the line, not the apartment — every unit in the affected column, plus the cellar, refuse and mechanical spaces at its base and any commercial tenancy. Second, monitors go into standardized positions in every one of those units and the counts are recorded and reported. Third, the structural scope is written down as an itemised list of penetrations by floor, so that it can be priced, scheduled and checked off rather than described as “sealing where necessary”. Fourth, access is treated as a project management problem with named responsibility, because it is the constraint that decides the outcome and it is nobody’s job by default.

The finish line is a data question. Sightings stopping is weak evidence, for the same reason sightings starting was weak evidence: a heavily reduced population is also an invisible one. What counts is zero across every trap position in the column, sustained over several consecutive service intervals — several, because egg cases carried at the time of treatment hatch afterwards and the emerging nymphs need time to encounter something. A single clean visit at the end of a long program proves considerably less than a board wants it to.

The last question is the one that decides whether the building is back here in three years. If the penetrations were sealed floor by floor, the riser has stopped being a corridor and the monitoring interval can lengthen honestly. If they were not, then the treatment reduced a population inside a structure that is still built to redistribute one, and the monitoring has to be permanent, because the condition that produced the first infestation is entirely intact.

That choice is the same one that runs under every subject in this cluster, and it is the same one facing a board dealing with bed bugs in the same building, along many of the same voids. The rest of the writing on risers, chases and service cores is indexed on the blog; the work itself is cockroach control; the species detail sits on the German cockroach profile; and the governance side — inspection scope, notice, records, who commissions common-element work — is written for boards and agents on the co-op and condo page.

If you manage a building where one line keeps coming back, the useful first step is not a treatment. It is a trap in every apartment in that column on the same day. The consultation is free and most of this can be worked out on the phone before anyone visits.

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