People have been talking about confits (cooked fat-preserved meats) -- either canned, locally produced commercial, or home-made -- especially for use in cassoulets. A question arose about safely "aging" confits that aren't canned. There's been related Chowhound discussion before, as in chowhound.chow.com/topics/572158, but I've attempted here to gather hard data from authoritative (mostly print) sources. Including ins and outs of a food poisoning hazard I first learned about in some depth long ago when I worked in a lab that researched it.
"Botulism" poisoning comes from food contaminated by toxic byproducts of growing _C. botulinum,_ one of the disreputable, ubiquitous, anaerobic, spore-borne Clostridium bacteria. Clostridia are everywhere, and are often innocuous, until they coincide with particular conditions to produce various ills. In foodborne botulism, the problem is not the bacterium but the toxin it makes.
This toxin is deadly but fragile. Cooking to 80 degrees C (176 F) for 30 minutes reportedly destroys any of it in food. Things are different with the tough, heat-resistant botulinum spores. The spores can survive hours of boiling and other common cooking methods. Treatment with steam pressure above 120 deg C (250 F) reliably kills them, and is mandatory in commercial canning. That protection is lost once a canned product is opened, or if a confit is made and used fresh. [Temperature/time details from Merck Manual, 2006 professional edition.]
Unless its spores are completely eliminated, botulinum growth and toxin can develop in food stored with little air contact ("anaerobic"). Contrary to some claims, complete oxygen exclusion is unnecessary for this growth [Merck]. Toxin can form at refrigerator temperatures, but not when frozen (which halts bacterial activity in general). Besides freezing, standard ways to discourage anaerobic bacteria are high acid levels in some foods (not meat confits) and strong chemical preservatives.
Current popularity of poultry confits has sometimes dangerously obscured the fact that originally, like other meats preserved for unrefrigerated storage, they were very highly salted, often with nitrite or nitrate preservatives ("saltpeter"). A detailed recipe in the old Larousse Gastronomique begins by steeping a cut-up goose in a kilogram (two pounds) of mixed salt and saltpeter, and claims indefinite shelf life for the result. In sharp contrast, as Harold McGee explains ["On Food and Cooking"], most modern non-canned meat confits are made to be eaten within a few days, therefore salted much more mildly, for flavor and color, not preservation. This "few days" also coincides with published guidance on other foods subject to anaerobic bacteria.
Botulism risk rises the longer these foods are kept without freezing. While I'm no expert, the cause of every US botulism food-poisoning case I've heard of was eating (without cooking further) foods either improperly canned, or stored out of air contact (such as under fats) for more than a few days. Possibly the most publicized case was among Berton Roueché's famous "Annals of Medicine." A Thanksgiving family reunion was poisoned (with a fatality) by mushrooms marinated by boiling in white wine for 30 minutes, draining, adding spices, covering in olive oil, and storing for two weeks. An investigating public-health doctor said that the remaining contaminated mushrooms gave no warning. "They looked good and smelled good. Delicious, in fact."
In a cassoulet or other stew, long further cooking protects against any botulinum toxin in the ingredients. The real danger would be if uncanned confit, aged without freezing, were then served, tasted, or just carelessly handled, without much further cooking. That's the risk to weigh against the alternative of aging it frozen, with some possible loss of texture. Luckily, unlike bacterial growth, the chemical evolutions desired for "aging," like other food chemical processes, continue (more slowly) at freezer temperatures.
Standard biochemical data put botulism in stark perspective. Here's a comparison of rough rodent lethal dosages, in milligrams (thousandths of a gram) per kilogram body weight. The smaller this number, the stronger the toxin:
Tubocurarine ("curare"): 30 mg/kg
Sodium cyanide: 15 mg/kg
A.-amanitin (mushroom toxin): 0.1 mg/kg
TTX (in Fugu fish): 0.01 mg/kg
Botulinum toxin: 0.0000003 mg/kg (0.3 nanograms/kg)
By this measure, pure bo-tox is some 100 _million_ times more poisonous than curare or cyanide. A little dab'll do you in! (Obviously, bo-tox used medically as a muscle relaxant is greatly diluted.) That's why carelessly storing foods subject to anaerobic bacteria is like Russian roulette: You may often win, but there's more to lose.