Bill Mollison Brought a Clipboard
The argument shows up at every table where permaculture people and aquaponics people end up sitting together, and it runs about the same way each time. Aquaponics needs pumps. It sits in tanks. It’s full of plastic. Where’s the soil? Where’s the succession? Where’s the part where the system runs itself after year seven?
The response from the aquaponics side is usually defensive — a list of the ways the system is almost permaculture, delivered in the tone of a man explaining why his truck counts as a farm vehicle. That concedes the frame before anybody’s said anything. It accepts that permaculture is a membership organization with a dress code, and that the job on the table is getting admitted.
There’s a better move available and it costs nothing. Stop asking whether aquaponics qualifies and start reading Mollison’s aquaculture work as what it is — a design brief that nobody finished.
Pollution is a relationship, not a substance
His definition is the load-bearing wall here, so it’s worth restating carefully. Pollution is an output of one element that no other element in the system is using. That’s a claim about relationships and placement rather than chemistry. Ammonia leaving a fish farm into a creek is pollution. The same ammonia arriving at the root zone of a head of lettuce is fertility. The molecule doesn’t change. The address does.
Hold that against what the two parent industries actually do. Recirculating aquaculture produces protein and a nitrogen-loaded effluent it pays to discharge. Hydroponics produces greens and buys nitrogen in a bag, then pays to discharge spent solution. One sector has a surplus it treats as a liability, the other has a deficit it fills by import. Couple them and both waste streams and both import lines cancel.
Hand that to Mollison as a napkin problem and the same drawing comes back. Which means the objection was never about the nutrient logic. It’s about the hardware.
The primitivism trap
Somewhere along the way a design science picked up a set of aesthetic loyalties. Hand tools over machines. Earth over polymer. Slow over fast. Those preferences came from somewhere real — a reaction against extractive industrial agriculture and against the assumption that every problem yields to more horsepower. Nobody needs to relitigate whether that reaction was earned.
But preferences harden into a purity test, and the test starts governing decisions it was not built to govern. You end up watching people decline to solve a problem because the solution arrived in a shape that offends. A gasket becomes a moral question.
That’s a strange fate for a body of work whose founding move was to look at a hillside and ask where the water goes.
Mollison wasn’t a primitivist. He was an ecologist with a bulldozer on speed dial, comfortable with earthworks at scale and with machinery when machinery was the efficient answer, and he carried a long-running impatience with people who mistook the trappings for the method. What he assembled was a way of asking questions about flows and functions. It travels into any technology you point it at, including technologies he never saw.
He wrote most of this already
The aquaculture chapter in the Designers’ Manual is one of the longest in the book, and he liked pointing out that a pond will outproduce the same area of pasture by a wide margin.
His worked examples tell you what he was after. The chinampas — raised beds built up from a shallow lake bottom, canals running between them, muck dredged from the canals thrown back onto the beds, fish and waterfowl living in the water column, several crops a year off the same ground for centuries. The rice-fish-duck-azolla systems, where the fish fertilize the rice, the ducks eat the pests, and the azolla fixes nitrogen on the surface between them.
Those are aquaponic systems running on mud instead of expanded shale and on a seasonal clock instead of an hourly one. The topology matches. Fish nutrient feeds plants, plants clean water, water returns.
What the chapter assumes throughout is land. Ponds sited by contour. A dam at a saddle. Swales between catchment and reservoir. Room to let the loop breathe across acreage.
Take the land away and the design problem gets interesting.
The same system with the acreage removed
Here’s what mine looks like. Three rooms: fish room, refugium and utility, greenhouse. Water leaves the fish tank, moves through a multi-tank refugium sequence, runs the greenhouse, and returns to a cistern in the fish room that gets refreshed with rainwater. The tank is the neutral point in the circuit, so that’s where the drawing starts.
Fish room and refugium run dark. Light exclusion is a nuisance to build and it settles the algae problem before it begins, which is the kind of trade Mollison made constantly — spend on structure once, stop paying maintenance forever.
The greenhouse runs an 1,100-gallon tank at one end and an 18-foot flowing basin, two feet wide, along the length. Potted plants sit partially submerged in the basin so the wicking does the irrigation. Crayfish live in the channel. That arrangement is a chinampa. Bed alongside canal, livestock in the water column, nutrient moving between them without anybody carrying a bucket. I didn’t set out to reproduce it and there it is, because the constraints that produced the original haven’t changed.
The planters are half rock, half soil, which comes from Will Allen. His insight was that running water through a system doesn’t require abandoning soil biology, and once you accept that, the media bed stops being an inert substrate and becomes habitat. Worms go in. Fungal networks establish. The bed starts producing soil as a byproduct rather than sitting there as a filter medium you eventually have to replace.
The roof is a steep arch because wet snow needs somewhere to go. A hardware mesh standoff runs three to four inches off the roof and walls. That standoff is espalier training infrastructure for cherry, plum, and mulberry in fan form, and it’s also a convection channel along the entire envelope, and it also keeps foliage off the glazing. One element, three functions, and none of them were the reason I put it there first.
Stocking runs across multiple tanks on a rotation, which is where the stability comes from. A single cohort coming out all at once would be a nitrogen cliff — the load that’s been feeding the plants walks out the door in an afternoon and the beds find out about it a week later. Staggering cohorts across tanks means no harvest is ever large enough against the whole to do that. Perch go in as fingerlings in spring, fast growers get culled and eaten through the year to hold the cannibalism down, and the bulk of a cohort comes out around month fourteen to sixteen, but the system’s total biomass never leaves at the same time. It’s relay cropping applied to fish. Nobody plants the whole garden in one week and then acts surprised about forty pounds of beans in August.
The cycle still has low points, and those get used instead of fought. Bed maintenance, crop rotation, replanting, anything that disturbs root mass — all of it gets scheduled at the low-nitrate windows, when the plants are between demand peaks and there’s slack in the system. That’s observation doing work that engineering would otherwise have to do. Read the cycle, put the disruptive tasks where the cycle already has room, and you don’t have to overbuild capacity just to ignore timing.
Chickens close the fertility loop on the composting side, rabbits contribute to it, worms are soil production infrastructure. Duckweed is the answer to the one input still coming from outside.
The dark rooms exhaust into the greenhouse, so CO2 from fish respiration and bacterial activity arrives where plants want it. That’s the pollution definition operating forward instead of as a complaint — an output with nowhere to go, given somewhere to go.
And the 1,700 gallons of aerated water circulating through all of it generates enough convective movement that conventional greenhouse ventilation becomes a solved problem I never had to solve. Thermal mass, humidity source, air handler, fire reserve, irrigation reserve, protein. Stack functions long enough and the system starts handing you things you didn’t design for.
Where the compression created new problems
Two of them, and they’re the reason Allen and Barber matter here other than as name-checks.
Mollison’s aquaculture never had to answer the soil question, because his water sat in ground. Compress the system onto a slab and the substrate question becomes live: what holds the roots, what holds the biology, what happens to the media in year ten. Allen answered it from the urban side, where the same constraint had already forced the issue.
Mollison also never had to answer for solids at density. A pond at low stocking rate handles its own settlement. Push the stocking up to what a small footprint requires and the solids load stops being a background process and becomes the thing that will kill you. Glynn Barber’s engineering on the waste conversion side is where that gets addressed — his ratios don’t make sense until you understand what the conversion unit is actually doing to the solids, and once you do, the number stops looking impossible.
Neither of those is a departure from the design brief. They’re what shows up when you take the brief seriously in a footprint Mollison wasn’t writing for.
The open questions, named
Feed. The sharpest one and the one that hurts. A system running on imported pellets is a nutrient import operation with a decorative loop drawn around it, and the field has spent too long changing the subject. Every point of feed produced on site moves the build toward the closed condition — duckweed and azolla on a side tank, black soldier fly on kitchen and garden waste, worm reproduction harvested as a yield, mulberry as green feed. Duckweed is my working answer. It isn’t finished.
Pumps. Real constraint, real solutions. Aeration fails before flow does, so a small battery on the blower buys more resilience per dollar than anything else available. Gravity-fed tiers cut head toward zero. Airlift moves water on air you’re already pushing. Mollison had no objection to moving water uphill, he just wanted it done by ram, wind, or panel rather than by a meter, and those options exist.
Plastic. The consistency argument settles this faster than any defense of the material. EPDM liner is a polymer. So is poly pipe on a swale outlet, greenhouse film, drip tape, shade cloth, a tree guard, and the tote catching rain off the barn roof. Standard kit across the movement, none of it controversial. The line being drawn isn’t about the material, it’s about which uses got grandfathered in. And a food-grade tote pulled out of a waste stream and given twenty years of service is the pollution definition running in the affirmative — an accumulating hazard redirected into use. Ferrocement, fiberglass over ply, and clay-lined basins are all there for anyone who wants them, which is where the chinampas started anyway.
What this settles
Somebody who wants to argue that a flowing basin with crayfish under wicking planters, fed by a dark refugium and stocked on a rotation that never lets the nitrogen fall out from under the beds, isn’t permaculture is welcome to make the case. They’ll have to explain which part.
The design science was the point. It works on ponds, on greenhouses, on wastewater plants, on supply chains, on anything with an input, an output, and a place where the two could be introduced. Treating it as a lifestyle package instead of a method throws away most of what Mollison built, in exchange for the satisfaction of being correct at a table.
He drew the aquaculture chapter and then ran out of book. Finishing it is the work, and I am on it.
