--- title: "I made garum for six months" author: "Bo Lindqvist (@ferment)" date: 2026-03-21T08:33:38.256Z updated: 2026-03-21T08:33:38.256Z canonical: "https://jot.place/@ferment/i-made-garum-for-six-months" description: "Roman fish sauce in a bucket in Malmo, six months, 20 percent salt. What Pliny and the Geoponica actually say, and how it lost a blind tasting to a 39 kronor bottle." tags: - "craft" - "fermentation" - "latin" - "primary-sources" - "recipes" - "rome" --- # I made garum for six months Sprat is cheap. That is really the whole origin of this: five kilos of whole skarpsill cost me less than lunch, and I had been reading Pliny in the dead hour between the morning bake and closing. On 4 June a 30 litre food-grade bucket went onto my balcony. On 26 November I strained it. In between I received a letter from the housing association and lost the goodwill of the woman on the floor below me. ## What the sources actually say Less than you would hope, and none of it is a recipe in the sense a baker means. Pliny is the most quoted and the least useful, because he is not explaining how to make it, he is being sniffy about it: > aliud etiamnum liquoris exquisiti genus, quod garum uocauere, intestinis piscium ceterisque quae abicienda essent sale maceratis, ut sit illa putrescentium sanies. Another kind of choice liquid, which they call garum, from fish intestines and the other parts that would have been thrown away, steeped in salt, so that it is the *sanies* of things rotting.[^1] I asked @marginalia about that last word, because every English version I found softened it into "liquor" or "fluid". Her answer: *sanies* is what runs out of an infected wound, thin and discoloured, neither blood nor pus. In month three I understood the choice completely. He also tells you the good stuff came from Carthago Spartaria, was made from mackerel, was called *garum sociorum*, and that around two *congii* of it changed hands for a thousand sesterces. The leftover sludge is *allec*, the brine run-off *muria*. A full product line, and no method. The method is in the Geoponica, a Byzantine farming compilation carrying much older material, which gives several ways of doing it. The first one is the one everybody uses: > a deep close-woven basket is inserted into the centre of the vessel containing these fish, and the garos flows into the basket. This, then, is how the liquamen is obtained by filtering through the basket; the residue makes alix. Entrails salted in a container. Small fish, salted the same way: sand smelt, small red mullet, mendole, anchovy, anything small enough. Left in the sun, stirred often.[^2] Sally Grainger, who has done more experimental work on this than anyone, puts the Geoponica ratio at 7 parts fish to 1 part salt by weight, which is 15% salt. The one surviving Latin recipe, from Gargilius Martialis, wants oily fish layered with dried herbs and salt "two fingers deep", stirred two or three times a day for seven days. Grainger reckons that text is corrupt, and the ratio it implies works out nearer 1 part fish to 3 parts salt, which is not a ferment, it is a salt block. Then Apicius, where liquamen turns up constantly, and where Book 1 has a fix for when your fish sauce has gone wrong: *Liquamen si odorem malum fecerit, vas inane inversum fumiga lauro et cupresso, et in hoc liquamen infunde ante ventilatum.*[^3] Fumigate an empty vessel upside down with bay and cypress, air the sauce, pour it in. There is real comfort in a two thousand year old instruction for deodorising your fish sauce. Theirs stank too. ## The salt is the only part I did not improvise Everything else here was guesswork. The salt was arithmetic, because the salt is the entire safety mechanism. There is no acid step, no heat step, nothing kills anything. You are relying on water activity and nothing else. The numbers that matter: below a water activity of 0.85 you fall outside the low-acid canned food regulations, because pathogens are not considered a live risk down there. *Clostridium botulinum* generally will not grow below 0.93. The organism that reaches furthest down is *Staphylococcus aureus*, which can manage 0.86 with oxygen. Saturated sodium chloride sits at 0.753, which is why it gets used as a humidity calibration standard. So you want a brine near the saturated end of that range, not hovering above 0.9 and hoping. What I loaded: - 5.0 kg whole ungutted skarpsill, straight from the counter, not rinsed - 1.2 kg of mackerel and herring guts and trimmings, free, because nobody wants them - 1240 g coarse sea salt That is 20% salt on total fish weight. Fish is roughly three quarters water, so the salt dissolves into about 4.65 kg of water: 1240 / (1240 + 4650) is 21% salt in the water phase. Modern commercial fish sauce runs nearer 25%, roughly 3 parts fish to 1 part salt, and about 20% is generally held to be the floor before too much of the nitrogen leaves as ammonia instead of staying as amino acids. Grainger's argument is that Southeast Asian producers could ferment faster at 15%, and that their higher salt is partly local taste and partly a real fear of botulism. I am not in Andalusia, I do not have three months of that sun, and I was not going to be the man who gave himself botulism for a blog post. So: 20%, on a bench scale, not a scoop. :::warning If you make this, weigh the salt as a percentage of the *total* wet weight of everything going in, guts included, and do not go under 20%. There is no smell test, no colour test and no bubble test that tells you a fish ferment is safe. The salt is the control and it is the only control. If you are not going to weigh it, do not do this. ::: Sweden's own fermented fish sits on the other branch of the family. Surströmming finishes around 9% salt and its barrels run at roughly 6 to 8%, against 15 to 20 for ordinary salted herring, because salt was scarce up here and low-salt methods got developed instead. Down there the work is done by halophilic anaerobes, *Haloanaerobium* among them. Different organisms, different smell, different product. Not weak garum. A separate animal. ## Six months in a bucket The Geoponica says sun. Swedish sun is a joke: Malmö in July averages a high of about 23C and drops to 13C overnight, and three weeks in the fish were still recognisably fish. So in early July I moved the bucket into the cupboard next to the bakery's boiler, which holds 28 to 30C, and it started properly. I stirred it every second day with a stainless paddle, because "stirring frequently" is one of the few explicit instructions in the sources and because the unstirred top layer was where I got a skin. | Month | Temp | Look | Smell | |---|---|---|---| | Jun | 15 to 22C | Fish still whole, brine risen over the top, cloudy | Sea, salt, faintly sweet, like a fishmonger's ice late in the afternoon | | Jul | 28 to 30C | Bellies split, bones showing through, liquid gone amber | Cat food, then ammonia over the top of it | | Aug | 28 to 30C | Brown sludge under a thin clear layer | The worst of it. Ammonia, wet dog, blocked drain. Pliny's word | | Sep | 28 to 30C | Liquid darkening towards strong tea | Ammonia backing off, something soy-like arriving underneath | | Oct | 27 to 29C | No recognisable fish. Bones, skin, scales | Nutty, like an old parmesan rind, with a spike of ammonia when stirred | | Nov | 27 to 29C | Clear dark red-brown over a settled paste | Savoury and mineral, close to soy sauce, a faint smoked edge | Month three is when Kerstin from the floor below rang my bell, having assumed something had died in the ventilation. I explained. She listened to the whole explanation, said "jaha", and went back inside. Two weeks later a letter arrived from the board about odour nuisance in communal spaces, which was fair, and which is roughly why Roman salting works were built where they were. Baelo Claudia, on the Bolonia cove in Cádiz province, is unusual precisely because its *cetariae* sat inside the town, and most did not. Its southern quarter could have held around 37 of them. And it was respectable money. The Pompeian who dominated the trade, Aulus Umbricius Scaurus, had a mosaic laid in his own house showing his product: *G(ari) F(los) SCOM(bri) SCAURI EX OFFICINA SCAURI*, the flower of Scaurus's mackerel garum, from Scaurus's own workshop. A man whose fortune came out of rotting mackerel put the label on the floor where guests walked over it. For what the rest of that town was like, @marginalia/the-walls-of-pompeii-were-mostly-people-being-annoying. ## Straining, and the honest part I did it the Geoponica way, having come that far: a woven basket pushed down into the middle of the bucket, then patience, then muslin, then a coffee filter. From 6.2 kg of fish and 1.24 kg of salt I got 1.9 litres of sauce and 2.6 kg of paste. The borrowed aquarium refractometer tops out at 100 ppt and mine sat off the top of the scale, which confirms nothing except that the arithmetic was probably right. Then a blind tasting against a 39 kronor bottle of Vietnamese nuoc mam, three of us, both diluted the same, on rice. Two out of three picked the bottle. Including me. Mine is saltier, sharper, more mineral, thinner in the mouth. One clean savoury hit and then it stops. The nuoc mam has a caramel roundness and a long finish that mine does not have anywhere, and there is no mystery in that: the bottle spent something like eighteen months to two years in wooden barrels, and mine spent six months in a plastic bucket, three weeks of it wasted on a balcony in Skåne. So I reproduced, for 210 kronor of fish and a letter from the housing association and half a year of my flat smelling like a wound, a product that sits on a shelf at the Asian grocer on Bergsgatan for the price of two coffees. I would do it again, differently. More viscera at the same salt, because Grainger is right that the gut enzymes do the liquefying and 1.2 kg was not enough for 5 kg of fish. A wide shallow vessel instead of a tall bucket. Straight into the warm cupboard on day one. What the six months bought me is that the sources stopped being quotations and turned into instructions. The basket makes sense. The stirring makes sense. *Sanies* is not Pliny being rude, it is Pliny being exact, and month three earns him the word. The allec, by the way, is better than the sauce. Thick, dark, tastes like anchovy paste with the volume turned up. Nobody believes me and nobody will try it. [^1]: Pliny, *Natural History* 31.93. The prices and the definitions of *allec* and *muria* are at 31.94 to 31.97. [^2]: Geoponica 20.46, the first of several methods given there. [^3]: Apicius, *De Re Coquinaria* 1.7, on correcting liquamen. The same passage tells you to fix an over-salty batch with a *sextarium* of honey.