Measuring pH: how every method works, and where each one fails

This is a guide to how the different ways of measuring pH actually work, what each is good and bad at, and which one suits which job. It is not a buyer’s guide and it does not crown a winner, because the honest answer is that the right method depends entirely on your water and what you are trying to learn from it.

Two ideas run through everything below, so they are worth stating up front.

What “pH” actually is

There are two definitions of pH, and the gap between them is the whole story.

The primary definition is metrological. It is fixed, by convention, using a cell that has no liquid junction — the Harned cell, a hydrogen-gas electrode paired with a silver/silver-chloride electrode, with the single-ion activity pinned down by the Bates–Guggenheim convention.^[1] This is run only in national metrology laboratories. It matters here for one reason: it proves that accurate pH is achievable — the limitation of everyday meters is not pH itself, it is the junction.

The operational definition is what your meter uses. You calibrate a glass electrode plus a reference electrode against standard buffers, and read pH by comparing millivolts.^[1] That reference electrode necessarily contains a liquid junction — and that junction is where most of the trouble lives.

So the number on your screen is a calibrated estimate built on an assumption: that the junction behaves the same in your sample as it did in the calibration buffer. In hard or buffered water, that assumption holds well. In soft or RO water, it quietly falls apart.

The two things that actually go wrong

Villain 1 — the liquid junction (an electrochemistry problem)

Where the reference electrode’s concentrated KCl leaks out to meet your sample, a small voltage develops because ions diffuse at different speeds (H⁺ and OH^- are unusually mobile). Calibration cancels most of it. The uncancelled remainder — the residual liquid junction potential (RLJP) — is the dominant systematic error in practical pH measurement, and it grows precisely when your sample’s ionic strength differs from the buffers’. That is exactly the situation in RO, soft, or rainwater.

The killer detail: two-point buffer calibration does not reveal this error, because all your buffers have similar (high) ionic strength — they agree with each other and lull you into confidence, then the dilute sample is where the bias appears. Illingworth’s classic study measured roughly 0.2 pH units of error per ten-fold difference in ionic strength, and found the fault present in 24 of 30 randomly chosen in-service electrodes across seven laboratories.^[2] A pooled metrology assessment put realistic combined uncertainty at about ±0.10–0.14 pH (k=2) for pH in poorly-buffered water over roughly pH 3.5–7.5.^[3]

Those tighter figures are matched-buffer specifications, valid in the calibration bottle, not in your RO tank. Dedicated low-ionic-strength recommendations exist precisely because of this.^[4]

Villain 2 — the water won’t hold still (a chemistry problem)

Soft water has almost no buffer intensity. For pure water the self-buffering term is β = 2.303·([H⁺] + [OH^-]),^[6][7] which is at its minimum near neutral and rises toward strongly acidic or alkaline. With so little buffering, a trace of dissolved CO₂ from the air swings the pH a lot, and it does so fast: CO₂ in- or out-gasses toward equilibrium the moment the sample contacts air whose CO₂ differs from the water’s.^[7] So a beaker of soft water is a moving target. Measure it sealed, in a flow cell, or at least quickly — an open sample can drift meaningfully while you watch. This is the same carbonate chemistry behind the KH/CO₂/pH triangle and why stable CO₂ is so hard to pin down in low-KH tanks.

The temperature trap

An electrode’s slope is the Nernst factor, 2.303·RT/F — about 59.16 mV/pH at 25 °C, and linear in absolute temperature (54.2 mV/pH at 0 °C, 61.5 at 37 °C).^[1] Two separate effects follow, and people conflate them:

• The instrument effect: a probe calibrated at one temperature and used at another mis-scales millivolts into pH. “Automatic temperature compensation” (ATC) corrects only this, and only away from the isopotential point near pH 7.
• The sample effect: the water’s real pH changes with temperature because the equilibrium constants do. ATC does not correct this. Best practice: measure near your calibration temperature and always record the temperature.

Every method, and where it fits

For each: how it works, what it is good at, where it fails, and its best-fit job.

1. The glass combination electrode (the default)

How: a thin pH-sensitive glass membrane develops a voltage proportional to pH, read against the built-in reference electrode.

Strengths: cheap, fast, direct, wide range (≈0–14), gives a live number you can act on.

Weaknesses: the RLJP problem above in soft water; the porous junction clogs and fouls; “alkaline (sodium) error” above ~pH 11; and it must be two-point (bracketing) calibrated with fresh buffers, because one point fixes only the offset, not the slope — and slope drifts as the electrode ages.

Best for: hard, buffered or tap water where ±0.1 is fine; and — importantly — tracking relative change (before versus after), where absolute accuracy matters less.

2. Better electrodes for soft water

Same principle, but engineered to tame the junction.

What changes: a free-diffusion, flowing, or ground-glass sleeve junction replaces the ceramic frit, continuously refreshing the salt bridge; ROSS-type electrodes swap the silver/silver-chloride internal for an iodide/iodine redox reference for extra stability; dedicated low-ionic-strength electrodes and ionic-liquid salt bridges push further.

Evidence it helps: in a direct comparison at low ionic strength, free-diffusion junctions outperformed ceramic plugs — expanded uncertainty at pH 7 of roughly 0.02 (capillary/free-diffusion) versus ~0.09 (ceramic).^[5]

Weaknesses: more electrolyte leakage, more fuss, and performance varies unit-to-unit. Even at their best, you are looking at perhaps ±0.05 pH with obsessive technique in soft water — better, not solved.

Best for: the determined hobbyist who genuinely needs a soft-water number and will maintain the kit.

3. ISFET and solid-state sensors

How: an ion-sensitive field-effect transistor (invented by Bergveld in 1970^[8]) senses H⁺ at a solid gate surface — no fragile glass bulb, low output impedance, fast, rugged.

Strengths: speed, robustness, stability, ideal for continuous logging.

The catch, and it is the important one: an ISFET is only half a cell. It still needs a reference electrode, and in normal use that reference still has a liquid junction — so the ISFET does not escape the low-ionic-strength junction problem. The famous oceanographic Durafet reaches ±0.005–0.01 pH,^[9][10] but it does so by pairing the ISFET with a junction-free chloride reference that only works because seawater has high, stable chloride. That trick does not transfer to freshwater. So: excellent sensor, same soft-water ceiling.

Best for: continuous monitoring and logging, especially in stable, higher-ionic water.

4. Colorimetric indicators (drop tests and the CO₂ drop checker)

How: a pH-sensitive dye changes colour across its transition range. The CO₂ drop checker is the clever one: bromothymol blue (pKa ≈ 7.1, yellow→green→blue over pH ~6.0–7.6) sits in a fixed 4 dKH reference solution behind an air gap, so only gaseous CO₂ crosses over — the colour becomes a proxy for dissolved CO₂ (green ≈ ~30 ppm), not tank pH. Separately, alkalinity (KH) titration drips acid to a bromocresol-green / methyl-red endpoint.^[19]

Strengths: junction-free (immune to the electrode’s biggest weakness), chemistry-anchored, cheap, hard to fool in a gross way.

Weaknesses: subjective colour reading, coarse resolution, and tannin/DOC-stained water masks the colour. And a specific gotcha: the titration endpoint is not a universal pH 4.5 — it rises as alkalinity falls (≈4.5 at ~500 mg/L, but ~4.9 at the ~30 mg/L of a soft tank, and higher still below that^[19]). Titrating rigidly to 4.5 over-titrates soft water and over-reads KH.

Best for: CO₂ tracking (drop checker) and robust KH/alkalinity — the pragmatic backbone for planted tanks.

5. Spectrophotometric (optical) pH — the reference method

How: add a purified indicator dye, measure absorbance at two wavelengths, and compute pH from the ratio using the dye’s laboratory-characterised pKa and absorptivities. No reference electrode, no junction, no buffer calibration.^[11][13]

Strengths: the three things that wreck electrodes are simply absent. In seawater it delivers precision ~±0.001 pH (1 SD) and bias <0.005^[13] and is the prescribed reference method for global ocean-acidification programmes.

Weaknesses / the honest caveats:

• It is range-limited by the dye — accurate only within ~±1 pH of its pKa. m-cresol purple (pKa ≈ 8.0 in seawater) suits pH 7–9, which is why ready-made ocean instruments do not read a pH-5 blackwater tank.
• The dye’s pKa shifts with ionic strength and temperature, so it must be re-characterised for your water — a seawater calibration is invalid in soft freshwater.
• The dye perturbs the sample, and that is worse in low-buffer water; rigorous work extrapolates to zero dye addition.
• Commercial dye is impure and biases the result low unless HPLC-purified.^[12]
• Freshwater maturity is uneven. Near-neutral freshwater is validated to ~±0.002 with purified, characterised dye;^[14][15][16] the acidic end is only emerging — demonstrated with bromocresol green in soil extracts and buffers^[17] — and no peer-reviewed study validates it in tannin-stained soft aquarium water (the colour breaks the baseline correction). That is an open gap, not a solved problem.

Best for: the gold-standard number if you will do genuine lab work with the right dye for your range — but there is no turnkey box for a blackwater tank.

6. Don’t measure pH — compute it (inference)

How: the carbonate system links alkalinity, dissolved CO₂/DIC, and pH through known equilibrium constants; measure any two and the third follows.^[7] Measuring alkalinity (KH) by titration and conductivity/TDS lets you calculate pH — junction-free, and using a titration that behaves well in exactly the soft water where electrodes fail. This is what the KH/CO₂/pH tables do.

Strengths: robust in near-neutral soft water; no probe, no junction; the whole chain is chemistry you can trust.

Weaknesses: it assumes carbonate alkalinity ≈ total alkalinity. In very low pH, near-zero-KH organic-acid blackwater, humic and fulvic acids contribute organic alkalinity, so measured alkalinity is no longer carbonate alkalinity and the proxy breaks down.^[18]

Best for: the everyday planted-tank question (“what is my pH, roughly, and is it stable?”) in soft near-neutral water — the method the science actually favours here.

Match the method to your water

There is no single winner. The table below is the short version; the picker beneath it walks you to the same answer from two questions about your own tank.

Your situation	The method that fits	Why
Hard / tap / buffered water, ±0.1 is fine	Cheap glass electrode	Junction error small; direct and fast
Tracking CO2 in a planted tank	Drop checker (+ electrode for change)	Measures CO2, not absolute pH
Absolute pH in soft/RO near neutral	Infer from KH + conductivity; or spectrophotometric (lab)	No electrode is trustworthy here
Blackwater at pH 4–5	A decent electrode is genuinely OK here	Abundant free H+ poises the reading; KH-proxy fails at ~0 KH
Lab-grade truth, cost no object	Spectrophotometric, dye characterised for your matrix	Junction-free reference method (±0.001–0.005)

The counter-intuitive line in that table is the last two: low pH is the easy regime. At pH 4–5 there is plenty of free H⁺ to give a stable, well-defined reading — which is exactly why hobbyists find all their meters agree and settle fast at low pH, and only diverge above pH ~6. The measurement crisis is not “soft water” in general; it is weakly-buffered water near neutral.

What the hobby often gets wrong

• “0.01 resolution = 0.01 accuracy.” No. Resolution is how many digits it shows; accuracy is how close it is. In soft water, realistic accuracy is ~±0.1 pH regardless of the display.^[3]
• “My cheap meter is consistent, so it is right.” Precision is not accuracy.^[20] A single-point-calibrated meter that reads fast and identically can be confidently biased — the very stability that feels reassuring can be the sign it is not showing the water’s real ambiguity.
• “The expensive meter is faulty because it wanders.” Often the opposite — in soft near-neutral water the pH genuinely is unstable, and the sensitive meter is honestly reporting it while the cheap one hides it behind a sticky reading.
• “Titrate KH to pH 4.5.” That over-titrates soft water; the true endpoint is ~4.9 (and higher) at low alkalinity, not a fixed 4.5.^[19]
• Honesty marker: the “hardest zone is near neutral” framing above is synthesis from buffer chemistry, not a stated conclusion of the metrology literature — treat it as reasoning.

The bottom line

Match the method to the water and the task. Absolute pH in soft, near-neutral water is hard for physical reasons that no amount of money fully fixes — and for a planted tank you rarely need it: infer it from alkalinity and conductivity. If you keep true blackwater at pH 4–5, the good news is that is the regime where a decent meter can actually be trusted. And if you want the real, defensible number at any cost, it exists — spectrophotometric pH with a dye characterised for your water — but it is a laboratory method, not a purchase.

References

1. Buck, R.P. et al. (2002). Measurement of pH. Definition, standards, and procedures (IUPAC Recommendations 2002). Pure and Applied Chemistry 74(11), 2169–2200. DOI:10.1351/pac200274112169.
2. Illingworth, J.A. (1981). A common source of error in pH measurements. Biochemical Journal 195(1), 259–262. PMID:7306051.
3. Kadis, R. & Leito, I. (2010). Evaluation of the residual liquid junction potential contribution to the uncertainty in pH measurement: a case study on low ionic strength natural waters. Analytica Chimica Acta 664(2), 129–135. DOI:10.1016/j.aca.2010.02.030.
4. Covington, A.K., Whalley, P.D. & Davison, W. (1985). Recommendations for the determination of pH in low ionic strength fresh waters. Pure and Applied Chemistry 57(6), 877–886. DOI:10.1351/pac198557060877.
5. Harbinson, T.R. & Davison, W. (1987). Performance of flowing and quiescent free-diffusion junctions in potentiometric measurements at low ionic strengths. Analytical Chemistry 59(20), 2450–2456. DOI:10.1021/ac00147a002.
6. Van Slyke, D.D. (1922). On the measurement of buffer values. Journal of Biological Chemistry 52(2), 525–570.
7. Stumm, W. & Morgan, J.J. (1996). Aquatic Chemistry, 3rd ed. Wiley.
8. Bergveld, P. (1970). Development of an ion-sensitive solid-state device for neurophysiological measurements. IEEE Transactions on Biomedical Engineering BME-17(1), 70–71. DOI:10.1109/TBME.1970.4502688.
9. Martz, T.R., Connery, J.G. & Johnson, K.S. (2010). Testing the Honeywell Durafet for seawater pH applications. Limnology and Oceanography: Methods 8, 172–184. DOI:10.4319/lom.2010.8.172.
10. Johnson, K.S. et al. (2016). Deep-Sea DuraFET: a pressure-tolerant pH sensor. Analytical Chemistry 88(6), 3249–3256. DOI:10.1021/acs.analchem.5b04653.
11. Clayton, T.D. & Byrne, R.H. (1993). Spectrophotometric seawater pH measurements with m-cresol purple. Deep-Sea Research I 40(10), 2115–2129. DOI:10.1016/0967-0637(93)90048-8.
12. Liu, X., Patsavas, M.C. & Byrne, R.H. (2011). Purification and characterization of m-cresol purple for spectrophotometric seawater pH. Environmental Science & Technology 45(11), 4862–4868. DOI:10.1021/es200665d.
13. Dickson, A.G., Sabine, C.L. & Christian, J.R. (eds.) (2007). Guide to Best Practices for Ocean CO₂ Measurements, SOP 6b. PICES Special Publication 3.
14. Yao, W. & Byrne, R.H. (2001). Spectrophotometric determination of freshwater pH using bromocresol purple and phenol red. Environmental Science & Technology 35(6), 1197–1201. DOI:10.1021/es001573e.
15. French, C.R. et al. (2002). Spectrophotometric pH measurements of freshwater. Analytica Chimica Acta 453(1), 13–20. DOI:10.1016/S0003-2670(01)01526-0.
16. Lai, C.-Z. et al. (2016). Spectrophotometric measurement of freshwater pH with purified m-cresol purple and phenol red. Limnology and Oceanography: Methods 14(12), 864–873. DOI:10.1002/lom3.10137.
17. Bargrizan, S., Smernik, R.J. & Mosley, L.M. (2018). A spectrophotometric method to determine pH in acidic (pH<5) soils. Talanta 186, 421–426. DOI:10.1016/j.talanta.2018.04.074.
18. Hunt, C.W. et al. (2025). Organic alkalinity and its application to carbonate system calculations. Limnology and Oceanography. DOI:10.1002/lno.12761.
19. APHA/AWWA/WEF. Standard Methods for the Examination of Water and Wastewater, Method 2320 B (Alkalinity).
20. JCGM 200:2012, International Vocabulary of Metrology (VIM), 3rd ed. (accuracy, trueness, precision).