How is blending water tested? The methods behind every number on a certificate of analysis

Key takeaways

• Conductivity is the only parameter cheap enough to watch continuously, which is why it anchors every purified-water standard. USP’s three-stage test starts with a deliberately uncompensated reading — Stage 1 limits run “from 0.6 µS/cm at 0°C to 3.1 µS/cm at 100°C” — before allowing compensation to 25 °C at Stage 2, where the limit is 2.1 µS/cm (Veolia).
• TDS and conductivity are not the same measurement, and a certificate can report either. The laboratory method is gravimetric — “filter through 0.45 µm, evaporate at 180°C, weigh residue” — while a meter reads conductivity and multiplies by a factor that varies with water type, from 0.50–0.55 to 0.65–0.70 (GAIMC).
• Organic carbon needs its own method. USP <643> caps Purified Water at 0.50 mg/L TOC, determined “by either subtracting the inorganic carbon (IC) from the total carbon (TC) or by purging the IC from the sample before oxidation” (Analytik Jena).
• Trace metals and anions are instrument work. EPA Method 200.8 determines 21 elements by ICP-MS at detection limits “1000-fold lower than ICP-OES” (Thermo Fisher); EPA Method 300.1 covers “bromide, chloride, fluoride, nitrate, nitrite, orthophosphate, and sulfate” by ion chromatography (US EPA).
• The most important test is the one no instrument performs. Flavour Profile Analysis uses “well-trained panelists who use specific, standardized terminology” (Standard Methods 2170) — the only screen that covers the odorous compounds a certificate never assayed.

Blending water is verified by a short list of standardised analytical methods, each matched to one class of contaminant: conductivity for dissolved ions, gravimetry or a conversion factor for total dissolved solids, catalytic or UV oxidation for organic carbon, ICP-MS and ion chromatography for individual metals and anions, culture on defined media for organisms, and a trained sensory panel for the faults that show up in a glass before they show up in an instrument. A certificate of analysis is a summary of those results for one sample, taken at one moment, by one laboratory. Reading it well means knowing what each method sees — and, more usefully, what it does not.

What does a blending-water certificate actually certify?

A batch, a sample, and a method. Not a process, and not the water in the tanker that reached you.

Every number carries three hidden qualifiers: which analytical method produced it, what that method’s detection limit was, and when the sample was drawn relative to when the water was filled. Two suppliers can issue certificates against the same demineralised water specification and describe materially different waters, without either being dishonest, simply because one reported calculated TDS from a handheld meter and the other gravimetric residue from an accredited laboratory.

That is not an argument against certificates. It is an argument for reading them as a set of methods rather than a set of numbers — and for asking the question a certificate structurally cannot answer: what happens between batches.

Why is conductivity the number that gets watched continuously?

Because it is the only parameter that can be measured in-line, in real time, for effectively no marginal cost — and because in very pure water almost everything else correlates with it.

The pharmacopoeial treatment is the most rigorous version of this test in commercial use, and its design is instructive. USP’s water conductivity chapter runs a three-stage procedure in which “Stage 1 is required for all samples, while Stages 2 and 3 are utilized only when a sample fails to meet Stage 1 test limits.” Stage 1 is “an online or offline non temperature compensated measurement” compared against a table whose limits run “from 0.6 µS/cm at 0°C to 3.1 µS/cm at 100°C” in five-degree increments. Only at Stage 2 may readings “be temperature-compensated to 25°C,” against a limit of 2.1 µS/cm; Stage 3 introduces pH-dependent limits “from 4.7 µS/cm at pH 5.0 to 4.6 µS/cm at pH 7.0” (Veolia).

Two details in that design matter to anyone buying water for spirit reduction.

The first is the refusal of temperature compensation at Stage 1. Compensation algorithms are built for water containing a normal salt load; applied to near-ionic-free water they introduce error rather than removing it. A certificate quoting conductivity without stating the measurement temperature has omitted half the result.

The second is why the test escalates at all. At these purity levels the dominant variable is not the plant — it is the air. Samples are “particularly susceptible to misreporting due to CO2 absorption and desorption effects” (Veolia). Atmospheric carbon dioxide dissolves into unbuffered water, forms carbonic acid, and raises conductivity while dropping pH — which is exactly the drift described in our note on reading pH and TDS specs. Stages 2 and 3 exist to distinguish a genuine ionic contaminant from a sample that simply met the room.

For reference, USP Purified Water pairs a conductivity limit of ≤1.3 µS/cm at 25 °C with ≤500 ppb TOC and a microbial action level of ≤100 CFU/mL (MECO) — three linked numbers on one certificate, which is the structure a blending-water sheet should imitate.

Why can two honest certificates disagree about TDS?

Because “TDS” names two different measurements, and only one of them weighs anything.

The reference method is gravimetric — Standard Methods 2540 C — and the procedure is literal: “filter through 0.45 µm, evaporate at 180°C, weigh residue” (GAIMC; Standard Methods 2540). It measures mass, it is slow, and at blending-water purity the residue can be smaller than the weighing error.

The routine method is conductivity multiplied by a factor. “Most handheld ‘TDS meters’ are actually conductivity meters with a fixed conversion factor (usually 0.5, 0.64, or 0.7) programmed inside. They measure EC and internally multiply by the chosen factor” (GAIMC). The correct factor is a property of the water, not the instrument: roughly 0.65–0.70 for natural freshwater, 0.50–0.55 where sodium chloride dominates, and 0.60–0.80 for mixed wastewater.

The gap between the two methods is the whole point. Conductivity sees charge; gravimetry sees mass. Anything uncharged is invisible to the first and fully counted by the second — the same asymmetry that governs which contaminants survive a reverse-osmosis membrane. A water carrying non-ionic organics reads clean on a meter and heavy on a balance, and the conversion “fails” precisely when “high organic content or non-ionic dissolved solids are present” (GAIMC).

Practical rule for a buyer: accept calculated TDS as a process-control figure, insist that the certificate state the conversion factor used, and treat conductivity in µS/cm at a stated temperature as the more honest number.

How is organic carbon measured — and why does sampling distort it?

By oxidising everything organic in the sample to carbon dioxide and measuring the result.

USP <643> is the reference procedure and caps Purified Water at 0.50 mg/L, with the value obtained “by either subtracting the inorganic carbon (IC) from the total carbon (TC) or by purging the IC from the sample before oxidation” (Analytik Jena). In a blending-water context TOC is the parameter with the most leverage per rupee: it is simultaneously the nutrient budget that decides whether a microbial population grows in storage and the proxy for the odorous small organics that a conductivity meter cannot see at all.

It is also the parameter most easily ruined by the act of testing it. Industry guidance on pharmaceutical water investigations notes the common practice of wiping a sample point with industrial alcohol before drawing a sample, and the consequence: “this alcohol is an organic material and this could result in a TOC excursion” (News-Medical). At single-digit parts per billion, the sampler, the valve, the bottle and the air are all plausible sources of the number being reported.

Which is the general lesson of low-level analysis, and worth stating plainly: at blending-water purity, a bad result is more often a sampling failure than a plant failure. That is an argument for in-line instruments where the water is made, and for treating a single out-of-trend certificate as a prompt to resample rather than to reject.

Which contaminants need instruments rather than meters?

The ones that matter at micrograms per litre and below — which is most of the ones that damage a spirit.

Trace metals: ICP-MS. EPA Method 200.8 “defines a procedure for the analysis of 21 elements in waters and wastes using ICP-MS,” with detection limits reaching parts per trillion, “1000-fold lower than ICP-OES” (Analytik Jena; Thermo Fisher). This is the method that answers the question a hardness figure cannot: how much calcium, magnesium, iron and copper are actually present. It is not an academic question — technical guidance for distillers holds that “as little as two ppm of calcium in the final product can cause precipitation” (Artisan Spirit), and iron is the contaminant most likely to arrive from pipework rather than from the plant.

Anions: ion chromatography. EPA Method 300.1 covers “bromide, chloride, fluoride, nitrate, nitrite, orthophosphate, and sulfate,” and was developed from its predecessor specifically “to meet a lower level of bromate (10 ppb)” (US EPA; Thermo Fisher). Bromate belongs on a blending-water certificate wherever ozone is used as the terminal disinfectant, and chloride belongs there as a corrosion indicator for stainless steel.

Organisms: culture on defined media. ISO 6222 enumerates heterotrophic colonies by pour plate at 22 °C and 36 °C; ISO 9308-1 detects coliforms and E. coli by membrane filtration onto chromogenic agar (Testing Lab). The incubation temperature and the growth medium are part of the result, not a formality — which is why a count reported without them is not comparable to anything.

What test catches the faults no instrument reports?

A panel of trained people, which remains the only broad-spectrum screen in water analysis.

Two standardised sensory methods exist. Threshold Odour Number is “a dilution-to-threshold test” and forms the basis of the US secondary standard for odour, though it “has been subject to criticism in the drinking water community for being a poor measure of consumer acceptance.” Flavour Profile Analysis is the more diagnostic successor: it “is designed to evaluate the tastes and odors important in drinking water,” and its strength is “its use of well-trained panelists who use specific, standardized terminology and can discern and express small differences between samples,” each assigning descriptors to what they detect (Standard Methods 2170; Standard Methods 2150).

The case for putting a sensory method on a blending-water programme is arithmetical rather than romantic, and we have made it at length in our note on tasting blending water directly. An instrument reports what it was asked to look for, and a certificate listing conductivity, TDS, pH, TOC and plate count is silent on geosmin, on 2-methylisoborneol, and on the chlorophenols that register in the tenths of a microgram per litre. A panel is silent on none of them — and it is the only check that catches a fault acquired after the certificate was issued, in a hose, a tank or a tanker.

Who has to test, and how often?

In India, the framework is prescriptive, and it is a licensing condition rather than a courtesy.

Under FSSAI’s scheme of testing for packaged drinking water, microbiological parameters are tested “once every month”; the chemical and general parameters — heavy metals, nitrates and nitrites, mineral oil, phenolic compounds, anionic surface active agents, calcium, magnesium, aluminium and borates — “once every three months”; pesticide residues “once every six months”; and all parameters prescribed under the regulation face “six monthly testing as a licensing condition,” with the frequency escalating to monthly after any failure until three consecutive compliant batches (FSSAI scheme of testing).

The limits themselves sit in IS 10500, which sets an acceptable TDS limit of 500 mg/L, turbidity of 1 NTU, pH of 6.5–8.5, total hardness of 200 mg/L and iron of 0.3 mg/L, each with a wider permissible limit in the absence of an alternate source (Auriga Research; IS 10500:2012). The corresponding test methods live in the IS 3025 series rather than in IS 10500 itself — the standard sets limits, the method standards say how to reach them.

Read that structure carefully and the gap is obvious: it is a potability regime, as covered in our guide to the regulatory standards for water used in spirits. A TDS acceptable limit of 500 mg/L is roughly two orders of magnitude looser than anything a proofing tank wants. Compliance is the floor. It is not the specification.

One more question belongs on the audit list, and it is about the laboratory rather than the water. ISO/IEC 17025 “specifies the general requirements for the competence, impartiality and consistent operation of calibration and testing laboratories,” and in India NABL accreditation is the formal recognition of technical competence against it (NABL India; ISO/IEC 17025). An in-house result and an accredited one are different objects. Both have their place; only one is evidence.

How should a trade buyer audit a certificate?

Seven questions, in order:

1. Conductivity in µS/cm, with the measurement temperature stated — and, ideally, a note on whether it was in-line or a grab sample.
2. TDS with its method named — gravimetric residue, or conductivity multiplied by a stated factor.
3. TOC in ppb, the parameter that predicts what the water will be like next week rather than what it was like on the day.
4. Trace metals by ICP-MS, with calcium, magnesium, iron and copper reported individually rather than folded into a hardness figure.
5. Anions by ion chromatography, including chloride, sulphate and — where ozone is used — bromate.
6. A microbiological count with the medium and incubation temperature named, plus the applicable pathogen-absence list and sample volume.
7. A sensory result, or an explicit statement that none was performed.

Then the two questions that decide whether the certificate means anything: is the laboratory accredited, and is this batch’s certificate or a representative one? As with the decision to make or buy blending water, the answers describe a supplier’s system. The numbers only ever describe one sample of one batch.

Glossary

• Certificate of analysis (COA) — a laboratory report of results for a specified sample of a specified batch. Evidence about that sample; inference about everything else.
• Conductivity — a water’s ability to carry electrical current, in µS/cm, driven by dissolved ions. The primary purity indicator for demineralised water, and meaningless without a stated temperature.
• Gravimetric TDS — total dissolved solids by mass: filter, evaporate, dry at 180 °C, weigh the residue (Standard Methods 2540 C). Its shortcut, calculated TDS, is conductivity multiplied by an empirical factor of roughly 0.5 to 0.7 depending on water type — fast, and only as good as the assumed factor.
• TOC (total organic carbon) — organic carbon measured by oxidation to CO₂. USP Purified Water caps it at 0.50 mg/L.
• ICP-MS — inductively coupled plasma mass spectrometry; the reference technique for individual trace metals at parts-per-trillion detection limits (EPA Method 200.8).
• Ion chromatography (IC) — separation and quantification of individual anions such as chloride, sulphate, nitrate and bromate (EPA Method 300.1).
• Flavour Profile Analysis (FPA) — a standardised sensory method using trained panellists and fixed descriptor vocabulary (Standard Methods 2170).
• ISO/IEC 17025 — the international standard for testing-laboratory competence and impartiality; accredited in India by NABL.

Frequently asked questions

What tests should a blending-water certificate include? At minimum: conductivity with temperature, TDS with its method, TOC, individual trace metals by ICP-MS, anions by ion chromatography, and a microbiological count with the medium and incubation temperature named. A sensory result should be present or its absence stated.

Why must conductivity be reported with a temperature? Because conductivity varies strongly with temperature and, in very pure water, standard compensation algorithms introduce error. USP’s Stage 1 test is deliberately “non temperature compensated” and compared against a table running from 0.6 µS/cm at 0 °C to 3.1 µS/cm at 100 °C (Veolia).

Can a laboratory test detect an off-flavour in water? Only if it was asked to look for that specific compound. Instrumental screens are targeted; a trained sensory panel using Flavour Profile Analysis is the only broad screen for unspecified odorous compounds (Standard Methods 2170).

How often is packaged water tested in India? Under FSSAI’s scheme, microbiological parameters monthly, chemical and general parameters quarterly, pesticide residues six-monthly, and all prescribed parameters comprehensively every six months as a licensing condition (FSSAI).

Does an accredited laboratory result matter more than an in-house one? For evidence, yes. ISO/IEC 17025 covers “the competence, impartiality and consistent operation” of testing laboratories (ISO/IEC 17025), and NABL accreditation is the Indian recognition of it. In-house testing is valuable for process control; accredited testing is what a buyer can rely on.

Is a drinking-water certificate good enough for blending water? No. IS 10500 accepts TDS up to 500 mg/L and iron up to 0.3 mg/L (Auriga Research) — limits set for safe consumption, not for a liquid that will be a third of a bottle of whisky. Potability is the floor beneath the specification, not the specification.

Blenders International · Water Science series