One water sample. Six pieces of kit. One defensible count.
Membrane filtration is only as good as its weakest link. This is the full walkthrough of the chain that carries a water sample from the bench to a colony count that survives an audit — what each component does, how to select it, and where recoveries quietly get lost.
For QC microbiologists and water-testing labs. Method-agnostic.
1. Everything between your sample and a count that survives an audit
Membrane filtration exists to solve a specific problem: in most waters, the organisms you care about are present at low densities in a large volume. You cannot plate a litre. So you pass that litre through a membrane with a pore size small enough to retain the target organisms on its surface, then transfer the membrane — organisms and all — onto a growth medium. Every viable cell that was suspended in the sample is now concentrated onto a countable surface.
That concentration step is the method's whole value, and also where it is fragile. A recovery is the fraction of viable target organisms in the original sample that end up as visible, countable colonies. Anything in the chain that stresses cells, retains them incompletely, lets them slip past, or carries them between samples pulls that recovery down — and a low recovery reported as a clean result is the failure mode auditors look for.
The six components of the chain. A complete filtration setup is not a parts list you assemble ad hoc; it is one specified system where each element is matched to the volume, the matrix, and the target:
The filtration base and funnel — holds the membrane, seals against the sample funnel, and directs filtrate to waste under vacuum.
The manifold — a 1-, 3-, or 6-branch stainless assembly that determines how many samples run in parallel.
The vacuum source — pulls sample through the membrane at a controlled, reproducible rate.
Disposable filter cups and syringe filters — eliminate between-sample carryover and handle small or pre-filtration volumes.
The gridded membrane — the retention surface; pore size and material decide what is captured and how cleanly it counts.
The media and incubation bench — selective/chromogenic media, colony counters, incubators, and reference strains that turn a loaded membrane into a defensible number.
The sections below follow the sample's own path — throughput, then carryover, then retention, then growth, then the validated whole.
2. Start with throughput
The manifold is the first thing to specify, because it sets your realistic daily capacity and the reproducibility of every filtration that follows it. Branch count is not a "bigger is better" decision — it is a match between how many samples you process per session and how much bench, vacuum, and hands you have to run them.
Configuration | Best-fit workload | Practical throughput | Trade-off to watch |
1-branch | Low sample counts, spot checks, method development, teaching benches | Serial, one membrane at a time | Simple and cheap, but becomes the bottleneck the moment volume rises |
3-branch | Routine QC labs with steady daily batches | Three samples in parallel per vacuum cycle | The common sweet spot; balances throughput against bench footprint and cost |
6-branch | High-volume commercial and municipal labs, peak-season F&B | Six in parallel; highest samples-per-hour | Demands adequate vacuum capacity and disciplined technique to keep every branch reproducible |
Controlling the filtration rate. Parallelism is only half of throughput; the other half is the flow rate through each membrane, and that is a recovery variable, not just a speed one. Excessive vacuum drives cells hard onto — and into — the membrane structure, physically stressing fragile organisms and, at the extreme, forcing small cells through the matrix. Too little vacuum lets the sample sit, which for some matrices means settling and uneven distribution across the membrane.
The goal is a controlled, reproducible differential across the membrane rather than the fastest possible pull. Stainless manifolds with a valve per branch let you bring each funnel to flow independently and shut it off cleanly the moment the sample clears, which is what keeps a 6-branch run as reproducible as a single filtration. Reproducibility beats speed: a run where every membrane saw the same differential and the same dwell time is a run you can defend.
3. No cross-contamination. No downtime.
Carryover is the quiet enemy of sequential testing. When one sample's organisms are transferred to the next through a shared, incompletely decontaminated funnel, you get false positives that are almost impossible to trace back — the contamination is real, it just belongs to the previous sample. In a workflow running many samples per session, the funnel is the single highest-risk shared surface.
Disposable filter cups. A disposable, pre-sterilised filter cup makes the sample-contact funnel single-use. Each sample gets a fresh, sterile path from funnel to membrane, so there is no shared surface to carry organisms forward and no flame-and-cool or autoclave cycle between samples. That removes both the carryover risk and the downtime that between-sample decontamination would otherwise impose — the "no downtime" is not a slogan, it is the eliminated re-sterilisation step.
Syringe filters. Not every sample suits a vacuum manifold. Small volumes, viscous or particulate-laden samples, and pre-filtration steps are handled by syringe filters — a membrane in an in-line housing driven by a syringe instead of vacuum.
Two typical jobs:
Small-volume filtration — where a full manifold setup is disproportionate to a few millilitres of sample.
Pre-filtration / clarification — removing gross particulates ahead of the analytical membrane so the retention surface isn't blinded by debris before it has captured the target.
The decision rule: if the funnel touches more than one sample, you have a carryover pathway, and single-use cups close it. If the sample won't behave under vacuum, move it to a syringe filter rather than fighting the manifold.
4. The membrane decides recovery
Of every component in the chain, the membrane has the most direct grip on your recovery. It is the retention surface: get the pore size, material, and surface finish right and viable target cells are held intact on a countable field; get them wrong and organisms are lost through the pores, damaged on capture, or grown into colonies you can't reliably count.
Pore size — retention vs. throughput. Pore size is chosen to reliably retain the target organism while letting sample pass at a workable rate. Too large and small cells slip through, understating your count; too small and the membrane blinds quickly with particulates and flow collapses. The pore size is matched to the smallest target you need to retain, not to the average.
Material family | General strengths | Typical consideration |
Mixed cellulose esters (MCE) | High porosity and even flow; long-standing default for many water bacterial counts | Broadly compatible with aqueous samples and colony transfer to media |
Cellulose nitrate | High protein/cell binding, clean colony development on the surface | Aqueous samples; avoid with solvents that attack the matrix |
PVDF | Low binding, high chemical resistance, robust and durable | Where chemical compatibility or ruggedness matters more than surface binding |
PES | Fast flow, low protein binding, high throughput | High-volume filtration where speed and low binding both matter |
PTFE | Maximum chemical and solvent resistance | Aggressive or solvent-based matrices rather than routine water counts |
Selection logic shown generally — confirm against your validated method.
Why the grid matters. A gridded membrane is printed with a fine, non-inhibitory grid that partitions the surface into equal cells. It does one job extremely well: it makes counting accurate and repeatable. The grid gives the analyst — or an automated colony counter — a fixed reference frame, so colonies are enumerated systematically across the whole surface rather than estimated, and two analysts counting the same membrane arrive at the same number. For any result that has to be defended, that repeatability is the difference between a count and a guess.
Sterile vs. non-sterile. Individually pre-sterilised, individually packed membranes remove a variable from the analytical path: there is no in-house sterilisation step to validate or to blame for a stray colony. Non-sterile membranes exist for non-analytical or pre-filtration roles where sterility isn't the point. For the analytical count, the pre-sterile membrane is one less thing between you and a clean negative control.
A selection order that works: fix the pore size to your smallest target first, then choose the material for matrix compatibility and colony development, then default to gridded plus individually sterile for anything you'll count and report.
5. Then grow it on the right media
A loaded membrane is potential, not a result. What converts it into a defensible number is the growth and enumeration bench: the right medium to make the target grow distinctly, controlled incubation to let it grow reproducibly, an accurate way to count what appears, and reference strains to prove the whole chain actually recovers what it's supposed to.
Selective and chromogenic media. Selective media suppress background flora so the target isn't overgrown; chromogenic media go further, producing a distinct colour reaction in target colonies via enzyme-substrate chemistry, so identification and enumeration happen in a single read. On a filtration membrane this pairing is what lets an analyst enumerate a specific organism directly from the surface — no ambiguous morphology calls, no confirmatory subculture just to know what to count. The membrane and the medium are chosen together: the medium is matched to the target the membrane pore size was selected to retain.
Incubators. Incubation is a controlled variable, not a shelf. Temperature stability and uniformity across the chamber directly affect whether colonies reach countable, characteristic size in the specified window — and different targets have different specified temperatures and times. Reproducible enumeration depends on every membrane in every run seeing the same, correct thermal environment.
Colony counters. Manual counting is slow and analyst-dependent; the gridded membrane already helps, and an automated or aided colony counter builds on it — imaging the membrane, counting systematically against the grid, and removing the drift between analysts and the fatigue of high-throughput sessions. The output is a count that's traceable and repeatable rather than a tally on a clicker.
Reference strains. Reference strains are how you prove the chain works. Running a characterised strain of known concentration through the same filtration-and-growth path validates recovery and confirms that media, incubation, and technique are performing — for method validation, for positive controls, and for the growth-promotion checks that keep a routine method honest. A count you can defend is one where you can show the same system recovers a known input.
6. One validated chain — not a parts list
Walk the chain back and the logic is a single thread. The manifold sets reproducible throughput. Single-use cups close the carryover pathway. The membrane's pore size, material, and grid decide what's retained and how cleanly it counts. The media, incubators, and counters turn retention into distinct, reproducible, traceable colonies. Reference strains prove the whole path recovers a known input.
Every one of those links multiplies into the final recovery. A perfect membrane behind an over-pulled vacuum still loses fragile cells; a validated medium behind a carryover-prone funnel still reports the previous sample; a flawless filtration read on a drifting incubator still under-counts. This is why the Water Filtration
Testing System is specified as one chain rather than sourced as independent parts — the components are matched to each other, to your volume, and to your target, so the recovery you validate is the recovery you keep run after run.
Six places recovery quietly gets lost:
Vacuum too aggressive — over-pulling stresses fragile cells and can force small organisms through the matrix, giving understated counts that look clean. Bring each branch to flow, shut off on clearing.
Shared funnel between samples — the highest-traffic carryover surface in the workflow. Single-use cups remove it entirely; re-used funnels demand a validated decontamination step every time.
Pore size set to the average target — size to the smallest organism you must retain, or it slips through and your count runs low. Retention is set by the hardest case, not the typical one.
Membrane and media chosen apart — pick them together. The medium must be matched to the target the pore size was selected to capture.
Incubator treated as a shelf — temperature drift and chamber non-uniformity move colonies out of the countable window. Every membrane in every run needs the same specified thermal environment.
No recovery evidence — without a reference strain run through the same path, you can't show the system recovers a known input, which is exactly what an audit asks you to demonstrate.
Specify the whole chain, not six separate purchases. Tell us your volume, matrix, and target organisms, and we'll match the manifold, membranes, media, and bench into one validated system — with the consumables thread that keeps it running.
Genesis Bioscientific — your complete microbiology solutions partner. Request a WFTS consultation · Explore the water testing range
