Three naming systems, one idea
Laboratories specify water by grade, and three specification families dominate. ASTM D1193 defines Types I through IV; ISO 3696 defines Grades 1 through 3; and CLSI's guideline defines Clinical Laboratory Reagent Water (CLRW) for medical labs. These documents are revised, and the grade names outlive the numbers behind them, so buy the edition your method or your accreditation actually cites and read the limits there rather than from a summary. They differ in detail but share the idea: match the water to the sensitivity of the work, because over-purifying wastes money and under-purifying wastes experiments.
The headline numbers, as published in the standards and summarised by the major lab-water vendors:
- ASTM Type I: resistivity above 18 MΩ·cm (in practice systems deliver 18.2), TOC below 50 µg/L — the grade for trace analysis, HPLC, cell culture media.
- ASTM Type II: above 1 MΩ·cm with low organics — general reagent preparation.
- ASTM Type III: above 4 MΩ·cm — rinsing, feed to Type I polishers, general lab use. (Yes, Type III is ionically purer than Type II; Type II's distinction is its low-organics heritage from distillation. The numbering is historical, not a purity ladder.)
- ISO 3696 Grades 1/2/3: conductivity at or below 0.1, 1 and 5 µS/cm respectively, with the standard's own caveat that it covers water for inorganic analysis — not organic trace work or biology.
- CLRW (medical laboratories): resistivity above 10 MΩ·cm, TOC below 500 µg/L, bacteria below 10 cfu/ml, with 0.22 micron final filtration.
Two cautions from the specifications themselves: ultrapure water does not keep — ISO 3696 says Grade 1 and 2 water should not be stored, because it pulls in CO2 and leaches from containers within hours — and "18.2 MΩ·cm" is a system capability, not a pharmacopoeial requirement.
The systems that make each grade
The engineering is a ladder, and published vendor practice (ELGA, Merck Millipore) is consistent about its rungs: reverse osmosis does the heavy lifting, removing 95 to 99% of ions and most organics; deionisation — mixed-bed cartridges or continuous EDI — takes RO permeate to Type II/III territory; and Type I is produced at the point of use by a polisher combining mixed-bed resin, dual-wavelength UV (185 nm to oxidise trace organics, 254 nm germicidal) and a 0.2 micron final filter. Nuclease- and pyrogen-sensitive work adds an ultrafilter at the dispense point.
For a South African lab the practical layout is usually: a small RO system with storage feeding the general lab ring, and benchtop polishers at the instruments that need Type I. Note what that ring is and is not. Reverse osmosis alone typically lands near the Type III resistivity figure on a good feed, but storage works against it — dissolved carbon dioxide and anything leaching from the tank both pull resistivity down, so a stored ring meets the grade only if it is verified at the point of use, on the parameters the method actually names. Do not describe a system as producing Type III because of what feeds it; describe it that way only when the water is measured there. Feeding polishers with RO water rather than tap water multiplies cartridge life; feeding instruments with stored "pure" water defeats the point.
Autoclaves, glassware washers and steam
Steam steriliser and washer manufacturers ask for purified feed for the same reason boilers do — scale — plus chamber corrosion from chloride. Published manufacturer guidance puts ideal autoclave feed in the low-microsiemens range, with RO-grade water sufficing for routine research loads; healthcare sterile processing in the US now formalises "critical water" for final rinse under AAMI ST108. The honest engineering answer is that requirements vary by machine: consult the manual, and expect RO or DI feed to be the recommendation.
Accreditation context
SANAS accredits South African testing laboratories to ISO/IEC 17025, which requires labs to control everything that affects result validity — reagent water included, via the water grades their methods specify. There is no SANAS-prescribed water grade; the analytical method drives the requirement, which is exactly how the grading systems are meant to be used.
Rule of thumb
Buy the grade the method demands, produce it as late as possible, and never store what you polished: reverse osmosis with storage for the building's general duty, verified where it is drawn, and point-of-use polishing for the instruments. If a lab is fighting baseline noise or blank contamination, water is the first suspect — and the cheapest one to fix.