Drinking water
The thing that kills first, and the one problem on this site that is completely solved — provided you do it every single day, including the day everyone is exhausted.
Home: Make water safe, Level 0 · Triage, level 1 of 3 in its track. Serves 4 steps.
Diarrhoeal disease is the largest killer in every collapse, every siege and every refugee camp on record. It outkills violence, cold and hunger, and it does it to the youngest and the oldest first. It is also, unlike almost everything else on this site, a solved problem: the fix is heat, heat is free, and it has been understood for a century and a half.
So the difficulty is never knowledge. It is discipline. Groups do not die because nobody knew to boil the water; they die because on day forty somebody was tired, the fire was out, the water looked clean, and it was only one cup.
This guide is built around that. It covers what to look at before you touch the water, how to make it safe, the three or four widely believed methods that do not work, and the hazards that boiling cannot touch at all. It ends with the part that actually decides the outcome: how to make the routine survive contact with a bad week.
Reading the water before you touch it
You will rarely get a choice between good water and perfect water. You will get a choice between several bad options, and the skill is ranking them fast.
Two rules do most of the work. Moving beats still, because still water grows things and concentrates what falls into it. Underground beats surface, because soil and gravel are a filter that has been running for longer than you have been alive.
But the ranking below is about how much treatment the water needs, not about whether you can skip treatment. There is exactly one entry on this page you can drink untreated, and even then only if you can see it leaving the rock.
What the water is telling you
These are the judgements you make standing on the bank, before any equipment comes out. The verdict is about how much work the water needs — not a licence to skip the boiling.
Clear it first, or nothing else works
Every disinfection method on the next page is defeated by cloudy water. Suspended particles physically shield microbes from heat, from chlorine and from ultraviolet, and organic matter eats chlorine before it reaches anything alive. Clarifying is not a cosmetic step and it is not optional.
The cheapest method is time. Fill a vessel, leave it, and the sediment falls out on its own. A three-pot arrangement makes this continuous: the newest water settles in the first, yesterday's decants into the second, and the clear water from the second goes to the fire.
For faster results, or where the particles are too fine to settle:
- Cloth, folded four to eight times. Tightly woven cotton is best; a folded sari removes a surprising amount, including the copepods that carry guinea worm. Always fold the same face down, and wash and sun-dry it between uses.
- A sand-and-gravel pre-filter. A bucket with gravel below and sand above, feeding the settling pots. This is not disinfection, and water out of it is not safe to drink.
- Coagulation. Alum, if you have it, at roughly a teaspoon to twenty litres — stir hard for a minute, slowly for five, then leave an hour. Crushed Moringa oleifera seed does the same job where the tree grows: about one seed per litre, ground, and it removes most turbidity and a good share of the bacteria with it.
What you are aiming for is water you can read newsprint through. That is not clean, and it is not safe. It is merely ready to be made safe.
The methods that work, ranked by how little can go wrong
Only four methods matter at small scale. They are ranked here by reliability under bad conditions — tired people, poor light, no measuring equipment — not by elegance.
Boiling is first and it is not close. It needs no dose, no contact time, no measurement and no working knowledge of chemistry. The water is safe when it reaches a rolling boil, and the killing is essentially finished on the way up. Holding it there for minutes is a myth that costs fuel; the extra minute per 1,000 m of altitude is margin, not requirement. Boiling handles everything living — bacteria, viruses, protozoal cysts including Cryptosporidium, which resists chlorine badly.
The reason to know the others is fuel. Boiling every litre a group drinks is a large and permanent firewood bill, and firewood is the constraint that eventually decides how long a camp can stay where it is.
| Method | What it needs | What it kills | What it misses | Cost |
|---|---|---|---|---|
| Boiling | Fuel and a pot. A rolling boil is the whole indicator | Everything living, including Cryptosporidium and viruses | Chemicals, metals, salt, cyanotoxins — and it concentrates all of them | Heavy and continuous fuel use |
| SODIS — sunlight | Clear PET bottles under 3 litres, full sun, 6 h (2 days if overcast). Water must already be clear | Bacteria, viruses, most protozoa | Poor against Cryptosporidium; fails in cloud, fails in turbid water, fails in a big or coloured bottle | Free, but slow and weather-dependent |
| Chlorine | Bleach or chlorine tablets, and clear water. Roughly 2 drops of unscented 5 % bleach per litre; wait 30 minutes; faint chlorine smell means it worked | Bacteria and viruses reliably | Cryptosporidium almost entirely. Consumed by organic matter before it works | Depends on a supply you cannot make |
| Slow sand filter | A barrel, graded sand, and weeks of patience while the biological layer forms | 90–99 % of bacteria and most protozoa once mature | Viruses, chemicals. Not a substitute for boiling drinking water | Build once, run forever, no fuel |
| Settling and cloth | Time and a folded cloth | Nothing | Everything — this is preparation, not treatment | Free |
Slow sand: the only method that scales
Boiling solves the problem one potful at a time and burns wood doing it. A slow sand filter solves it continuously, for a household or a settlement, using no fuel at all — and it is old, forgiving technology that predates any understanding of why it works.
What makes it work is not the sand. It is the schmutzdecke, a living biological layer that forms on the top few millimetres over the first few weeks of running, in which organisms consume what arrives. A brand-new filter is a sieve and removes very little. The same filter a month later removes most bacteria and most protozoa.
Three things follow from that, and all three are counter-intuitive:
- The filter must never dry out or be starved of water. The layer dies, and you are back to a sieve with no warning that anything changed.
- Slow is not a limitation, it is the mechanism. Running it fast pushes water past the biology. Roughly 0.1–0.3 m per hour of face velocity is the working range.
- Cleaning means skimming the top centimetre, not washing the sand. Wash it and you have thrown away the part that does the work, and you start the weeks again.
A mature filter is excellent at bacteria and protozoa, mediocre at viruses, and does nothing at all to dissolved chemicals. In practice: filter everything, and boil what you drink.
What does not work — and what it costs to believe it
Everything in this section is widely believed and regularly repeated. Each one has killed people who thought they had treated their water.
- "Running water is safe." It is better than standing water, and that is all. A fast, clear mountain stream carries Giardia perfectly well, generally from an animal you cannot see a kilometre upstream. Clarity is a statement about sediment, not about biology.
- "You can tell by looking, smelling or tasting." You cannot. The infectious dose of some organisms is under ten cells. No water on earth looks like ten cells.
- "Charcoal filters it." Covered above. Taste and some chemicals, not pathogens.
- "Cloth or a sand filter is enough on its own." Both are preparation. Neither disinfects, and a fresh sand filter barely even reduces.
- "Boiling for ten minutes is needed." No — reaching a rolling boil is sufficient, and the myth burns fuel that a group cannot spare, which in practice means people boil less often rather than longer.
- "Alcohol in it makes it safe." No. The concentration required is far above anything drinkable, and the belief has killed people who diluted spirits with river water.
- "A little is fine." The single most common failure. Infection is not proportional to volume.
- "It is fine, we have been drinking it for weeks." Contamination is an event, not a property. The latrine seeps, the animal dies upstream, the flood comes, and yesterday's safe source is today's outbreak.
The hazards heat cannot touch
Everything above this section is about things that are alive. This section is about the ones that are not, and the rule for all of them is the same: boiling does nothing, filtering through sand does nothing, and reducing the volume makes it worse.
These are slow poisons. Nobody notices them in week one, which is exactly why they belong in a guide read in week one — a source chosen now may be drunk from for years.
| Hazard | Where it comes from | What it does | What actually helps |
|---|---|---|---|
| Arsenic | Natural in some groundwater — notoriously in the Ganges–Brahmaputra delta. Odourless, tasteless, invisible | Skin lesions, then cancers, over years. Affects tens of millions of people | A different source. Deep wells vary hugely well to well; iron-based filtration and coagulation work but need testing to confirm |
| Fluoride | Volcanic and granite groundwater — the East African Rift, parts of India and China | Mottled teeth, then crippling bone disease at high doses | Blend with low-fluoride water, or bone-char filtration. Boiling concentrates it |
| Nitrate | Latrines, manure and fertiliser leaching into shallow wells | Blue-baby syndrome — potentially fatal in infants under six months | Move the well or move the latrine. This one is usually self-inflicted and preventable |
| Lead | Lead pipe, lead solder, lead-glazed pots, some salvaged plumbing | Cumulative neurological damage, worst in children | Do not use lead anywhere in the water path. Never draw drinking water from the hot tap of an old system |
| Salinity | Coastal aquifers, over-extracted wells, storm surge | Undrinkable above about 1 g/l; seawater actively dehydrates | Only distillation or a solar still. No filter removes salt |
| Cyanotoxins | Algal blooms, as above | Liver and nerve damage; kills livestock and dogs quickly | Avoid the water entirely. Heat-stable and not removed by sand |
Carrying and storing it without putting the contamination back
A remarkable share of waterborne disease is contracted after treatment. The water was boiled properly, and then it was poured into the vessel it came from, or a hand went in it, or the cup that was used to serve it had been sitting on the ground.
The rules are short and they are all about hands:
- Narrow necks beat wide mouths. A vessel you cannot get a hand into cannot be contaminated by a hand. This is why the jerrycan and the narrow-mouthed pot exist in every culture that carries water.
- Pour, never dip. One dedicated dipper with a long handle, hung up and never laid down, if you must dip at all.
- Keep two vessels and never mix their roles. One is raw, one is treated. Mark them so unmistakably that a tired person in the dark cannot confuse them. Never let the raw vessel touch the treated one's lip.
- Cover everything. Flies and dust are the two commonest re-contamination routes after hands.
- Store it cool and dark, and turn it over. Treated water is not sterile and does not stay good indefinitely. Boil what you will drink in a day rather than accumulating a week's worth.
- Wash and sun the vessels. Ultraviolet is free and does most of the work on the inside of a rinsed container left open to the sky.
Making the routine survive a bad week
This section is the reason the guide exists. Nothing above is difficult; all of it is done correctly for the first month and then decays. What follows is how groups that survived kept it going.
- One person owns water, by name. Not "everyone is responsible", which means nobody is. The role rotates on a schedule, and the handover is spoken aloud.
- The treated vessel is physically different. Different shape, different colour, different place. Never a matter of remembering which is which.
- Boil on a schedule, not on demand. A fixed potful at a fixed time, every day, whether or not it seems needed. Treatment that happens when someone is thirsty is treatment that gets skipped when someone is busy.
- Site for it once, properly. Camp above and upstream of the latrine, 30 m minimum, and check where the water goes when it floods rather than where it sits today. This decision cannot be revisited cheaply.
- Watch for the first case and act on it immediately. One person with watery diarrhoea is an event to investigate — which source, which vessel, which meal. Two is an outbreak, and by then the cause is a week old.
- Teach it to the children explicitly. They will drink from the stream on the way back from anywhere unless the rule has been said out loud to them, repeatedly, in specific terms.
Build the gear
Where to go next
This page covers what you are most likely to meet. It is not a field guide, and on identification that difference can kill you. These are the works that go further — chosen because they can be held offline.
Still to come
Illustrations carry a technique further than prose, but some things only move well. Recorded here rather than left implicit.
- Figures for the spring box and the sealed well head — the two constructions that turn a caution-rated source into a good one
- A solar still figure, for the salinity case, which is currently the only hazard here with no drawn answer
- Photographs of a working slow sand filter's schmutzdecke, mature and freshly skimmed
- Regional guidance on arsenic and fluoride belts — where the known problem aquifers are, which is knowledge no amount of on-site testing replaces