Testing and Measurement

A number can go wrong in four places: the sample (the wrong water, the wrong amount, or water left sitting), the tool (dirty, expired or not calibrated), the reading (read too early, misread, or in the wrong unit) and the record (the wrong tank, the wrong box, or a typo).

When a number looks strange, don't ask "what's wrong with the tank?" first. Ask "what could be wrong with the number?" Check the sample, the tool, the reading and the record. Most strange numbers are test errors.

Accuracy is how close a result is to the true value. Precision is how closely repeated tests of the same water agree with each other. A test can be precise and still inaccurate: it gives nearly the same wrong number every time. Resolution is the smallest step a test shows, like 0.1 dKH or 0.01 ppm. A display showing 0.01 steps isn't promising the result is right to 0.01.

Makers state accuracy as a range. One marine alkalinity checker states its accuracy as 0.3 dKH plus 5% of the reading. So when it reads 8.0 dKH, the true value could be anywhere from about 7.3 to 8.7. One reading is a range, not an exact number.

That's why a single reading isn't acted on alone. Compare it with earlier readings from the same tool and method, and retest any big move before writing it down (Module 6).

The detection limit is the lowest amount a test can reliably tell apart from zero. One low-range phosphate checker shows results to 0.01 ppm, with a stated accuracy of 0.02 ppm plus 5% of the reading. A reading of 0.00 on that test means "less than about 0.02", not "no phosphate".

This matters because zero nitrate or phosphate isn't normal (Module 6). A reading at or below the detection limit is reported as low, the same as a reading of zero. Some lab reports print 0.00 for anything below their limit; "not detected" would be the more honest label.

Take the sample from well-mixed water in the main tank, in a clean container, and test it fresh. Water right at a dosing line tip or a top-off outlet hasn't mixed with the rest of the system yet, so it reads higher or lower than the tank really is. Water left sitting before testing can change, so test right away.

Take it at about the same time of day as the readings you'll compare it with. pH and some other readings change through the day (Module 6).

A titration kit adds a reagent of known strength to a measured sample, a little at a time, until the color changes. That color change is the endpoint. How much reagent it took gives the result. Alkalinity kits commonly work this way. Other kits add a fixed amount of reagent and match the color to a printed chart.

Both depend on people judging color, and people see color differently, especially in poor or colored light. Read the color against a white background under ordinary white room light, not under the blue aquarium lights. In one side-by-side comparison of alkalinity kits on the same water, results ranged from about 10% low to about 30% high, depending on the brand. So compare readings from the same kit, and if you change kits, test with both for a while to see how they compare.

Never pour unused reagent back into its bottle: it can contaminate the whole bottle. Reagent loses strength with age, so check the date.

The meter shines light through the vial of reacted sample and measures how much gets through. Anything else in the light path changes the number: fingerprints, water spots, bubbles, scratches, stains, or outside light getting in.

Hold vials by the top and wipe the outside clean and dry before each reading. Tap off bubbles. Put the vial in the same way every time, close the lid fully, and keep the meter's well free of dust. After each test, empty and rinse the vial; reagent left sitting in a vial can stain the glass for good. When a reading is unexpected, check the meter with the maker's calibration set of known values.

Most test errors come from the same few things: the wrong test chosen on the meter, the wrong sample size, reagent not fully added or mixed, reading before the reaction time is up, or expired reagent. Each one gives a number that looks real but isn't.

The standard measuring device for liquids is a 50 mL graduated cup with a separate 1 mL section, such as the one that comes with Red Sea Reef Energy AB+. It's one cup with two measuring areas: the main area, marked in 5 mL steps up to 50 mL, and a small compartment at the back, marked in 1 mL steps up to 5 mL.

Use the small rear compartment for amounts of 5 mL or less, and the main area for anything larger. Small amounts can't be read accurately on the big lines. Set the cup on a flat surface and read it at eye level, with the liquid at the line. Use a clean cup, and rinse it in RO water after use, so one liquid doesn't carry into the next.

Use the same standard cup every time, so a measured amount means the same thing whoever measures it. A dose measured in a different cup or by guessing can't be compared or repeated.

Calibrating means setting a tool against a solution of known value, so it learns what that value looks like. Verifying means checking the tool afterwards, by reading a fresh solution of known value and seeing if the number matches.

Calibration solutions, called standards (or buffers for pH), are made to an exact value. They only work fresh. Pour out only what you need into a clean container, use it once, and never pour it back into the bottle.

pH probes are calibrated with at least two buffers, usually one to three pH units apart, for example 7.00 and 10.01. Rinse the probe in RO water between buffers so one doesn't carry into the next. Some meters show a slope after calibrating; 92% to 102% is acceptable. Outside that range, don't trust the probe: report it.

Calibrate and verify on the schedule your facility or the maker sets, and any time a reading can't be trusted. A second device checked against a standard can also verify a reading: if two trusted devices agree, the number is real.

A conductivity meter reads salinity from how well the water carries electricity. Calibrate it with a 35 ppt standard: put the probe fully in, tap off bubbles and let it finish. Rinse it in RO water before measuring anything.

A refractometer reads salinity from how much the water bends light through a prism. Calibrating it with 35 ppt calibration fluid is generally preferred over RO water, because refractometer scales aren't perfectly even across their range, so setting it near reef salinity is more accurate there. Use RO water for rinsing the prism. Give the sample about 30 seconds on the prism to reach the prism's temperature, and don't read right after carrying the refractometer between a cold room and a hot one. Models with ATC (automatic temperature compensation) correct for temperature, but only once the sample and prism have settled.

Salinity changes slowly, mostly with evaporation and top-off (Module 5). A big jump from yesterday is almost always the measurement, not the tank.

A temperature controller switches a heater or chiller on and off based on what its probe reads. If the probe comes out of the water, it reads the air and the controller heats or cools the tank the wrong way, while its display looks normal. Next to a heater, the probe reads the warm water around the heater, so the heater switches off early and the tank runs cold.

pH probes are stored the way their maker says, usually capped with storage solution. A probe left to dry out can stop reading accurately. If a pH pen reads a little off a reference, its offset is written on a label; apply it yourself before recording (Module 6). ORP probes get dirty and drift, and need cleaning and calibrating too (Module 6).

An automated tester takes water from the system, runs a test with its own reagents, and logs the result. For example, the Neptune Trident tests alkalinity at least four times a day and calcium and magnesium at least twice a day. Its maker states a precision of 0.05 dKH for alkalinity. Testers like this need reagent refills, a container for their waste, and calibration after the reagents are replaced.

Many readings a day show how a number moves through the day and over a week, which a single daily test can't. But an automated tester fails like any tool: reagents run out or go old, tubes clog, calibration slips. Check it against a trusted hand test from time to time, and treat a sudden jump in its readings like any strange number.

ICP testing (ICP-OES is the common type) measures many elements at once from one water sample. One lab's top package lists 43 parameters, including major elements, trace elements like iodine, strontium and iron, nutrients, and contaminants like copper and lead. The sample is collected in the lab's vials and mailed in. Results take about one to two weeks.

Lab results aren't automatically right. In one independent check, a reef lab tested a certified seawater standard of known values. Its results repeated well, and strontium was within about 1%, but manganese read about five times the true value. The authors concluded such results can show trends and large excesses or shortages, but may not be good enough to fine-tune trace elements.

Use lab tests for what hand tests can't do: trace elements and contaminants, checked now and then. Keep using hand tests for the daily and weekly numbers.

This is process of elimination. First rule out the reading: test again carefully, with fresh sample and the full method. If it's still strange, rule out the tool: verify it against a standard, or test the same water with a second trusted device. If the tool checks out and the number holds, it's real: record it and follow the form's ranges. If the tool is off, clean it, calibrate it, verify it, and test again.

When two different tests disagree on the same water, at least one is wrong. Don't average them. Check each tool against a standard, and trust the one that passes.

Test most often what can change fast and what the animals are most sensitive to. A system with many calcifiers uses alkalinity every day (Module 6), so it needs alkalinity tested more often than a fish-only system. Things that change slowly can be tested less often. Trace elements and contaminants go to a lab now and then.

Test more after any change: a new dosing setting, new animals, new equipment, or a change of salt mix. That's when numbers move.

Keep the method fixed: the same tool, the same method, at about the same time of day. Changing any of those changes the numbers, and then nobody can tell a real change from a different tool. Put calibration and verification on the schedule too.

Log one tank at a time, with that tank's form open. Use the unit and number of decimals the form asks for. Apply any offset label before you enter the number. If you tested at an unusual time, write the time down, so the reading isn't compared as if it were taken at the usual hour. Write what you saw and what you checked, so the next person knows what was already ruled out.

The log turns single readings into trends. One reading far off a steady line is most likely a test error: retest it. A slow creep in one direction, where each reading looks fine on its own, is drift, and only the log shows it (Module 6).