Bone stock

Start with the word, because it turns out to matter more than a terminology quibble should. Stock is bones and connective tissue simmered for hours — the thing that sets to jelly in the fridge. Broth is meat simmered briefly and seasoned to drink. Almost everything sold and studied as "bone broth" is a stock, and the two ends of that spectrum differ in protein content by a factor of seventy in the published literature. When a study reports that broth is a poor source of collagen precursors, the first question is which one it made.

The claim being tested is mechanistic: long-simmered bones and connective tissue release collagen, gelatin, glycine, proline, and glycosaminoglycans — raw material for cartilage matrix. The credibility arrives partly on loan from collagen-peptide supplements, which do have randomized trials behind them.

An earlier version of this entry concluded the chemistry refuted that. It doesn't, and the error is worth stating plainly because it ran the wrong way: this entry reported that a 300 mL serve of bone broth was the weakest source tested in a human feeding study, when the paper's results say it produced the highest total amino-acid exposure of any arm in that study. What the evidence actually lacks is an outcome, not a mechanism.

What's actually in the pot

A standardized amino acid analysis compared commercial and home-prepared bone broths against a 20 g reference dose of collagen supplement — the kind of dose used in the collagen research. Broth made to the standardized recipe came in significantly lower on glycine, proline, hydroxyproline, and hydroxylysine. But the finding is about preparation, not about the food: the non-standardized recipes varied enormously, and the café-prepared varieties had the highest levels of every amino acid measured, with self-prepared beating commercial. What the authors concluded is that broth cannot reliably deliver a known dose — not that it cannot deliver one.

The mineral picture is genuinely weak, with a caveat the marketing and the debunking both miss. A 2017 analysis found calcium and magnesium under 5 percent of daily recommended intake per serving. The same study also found that acidity and time move the number a great deal: dropping the pH from 8.38 to 5.32 raised calcium and magnesium extraction 17-fold and 15-fold, and cooking past eight hours extracted significantly more than shorter simmers. A long, vinegar-acidified stock is at the top of that range — which is still under 5 percent of daily intake.

The human data cut the other way, and the earlier version of this entry had them backwards. In an acute feeding study, a 300 mL serve of chef-made bone broth matched the 20 g collagen supplements on peak blood glycine (666 versus 646–747 µmol/L) and proline, and produced the highest total amino-acid exposure of any arm tested — significantly higher than both dairy proteins. It was the slowest to peak, not the weakest: glycine took 88 minutes against 48 for collagen peptides, and blood levels were still elevated when the measurement window closed at three hours. That broth turned out to contain 52.2 g of protein per 300 mL, including 11.8 g of glycine — more than the supplements it was being compared against.

What two days of simmering actually does

Miguel's stock runs 48 hours, which is the obvious lever to pull if you want more collagen out of the bones. The food-science literature says it doesn't work that way, and this is the least convenient finding on the page.

Collagen extraction plateaus early and then goes backwards. A chicken-feet gelatin study found no significant yield growth beyond about 2.5 hours of extraction. In a response-surface study of chicken-feet collagen, the fitted model's coefficient for extraction time is negative — longer runs recovered less, with the optimum at one to three hours. A chicken-skeleton study found total nutrient migration into the soup peaked at 180 minutes and gained nothing by 210. Prolonged high-temperature cooking also oxidises what is already in the pot.

Minerals behave the opposite way: they keep climbing with time and with acid, which is why the long acidified simmer is not pointless — it is just doing something other than what it is usually credited with. The honest summary is that hours three through forty-eight are extracting minerals and gelatinising what collagen came out early, not adding more.

The other thing prolonged simmering does not do is hydrolyse gelatin into supplement-sized peptides. After six hours at 121 °C — hotter than any stockpot reaches — roughly four-fifths of the protein is still above 10 kDa. Chicken feet held at 100 °C for eight hours still needed an added enzyme to get down to about 5 kDa. Home stock is gelatin, not collagen peptides. Whether that matters is genuinely unsettled: one crossover study found intact gelatin raised blood Pro-Hyp and Hyp-Gly nearly as well as a hydrolysate did.

The human evidence for joints

None. No trial, no cohort, no case series has ever measured a joint, cartilage, or pain outcome after bone stock. That is the ceiling on this entry and nothing below changes it.

What exists instead is a mechanism with most of its links measured. Gelatin is digested to hydroxyproline-containing dipeptides, chiefly Pro-Hyp, which appear in human blood within one to two hours of eating. In rats given radiolabelled Pro-Hyp, the label reaches articular cartilage and is taken up by chondrocytes and synovial cells — about 5 percent of what arrives in cartilage is still the intact dipeptide. And Pro-Hyp does something once it is there: in mouse chondrocytes it tripled glycosaminoglycan staining and roughly doubled aggrecan expression, while preventing cartilage thinning in a phosphorus-loading model.

The sharpest detail in that work cuts against the popular version of the theory. A mixture of free proline, hydroxyproline and the tripeptide Pro-Hyp-Gly did nothing at all. The dipeptide is the active species — which means "bone stock supplies amino-acid building blocks" is probably the wrong account of why it might work, if it works.

The animal literature is no longer a single study, as this entry previously said. Prophylactic bone broth reduced colonic inflammatory cytokines in mice with induced ulcerative colitis — a gut model that says nothing about cartilage. But a 2024 study fractionated chicken-vegetable bone broth, identified hyaluronan and chondroitin sulfate as the active constituents, and showed both the whole broth and that fraction raised bone mineral density in ovariectomised rats. Bone, not cartilage — but it is a skeletal outcome from bone stock, and it identifies which ingredients did it.

The lead question

A 2013 analysis found that broth made from organic chicken skin and cartilage carried 9.5 micrograms of lead per litre, versus 0.89 in the cooking water — and skin-and-cartilage preparations are exactly the ones promoted for joints. A larger 2017 study across multiple broth types measured only a few micrograms of lead and cadmium per serving and judged the ingestion risk minimal. The better-controlled data are reassuring, so the alarm is probably overstated for normal consumption, but the original signal has not been formally replicated and resolved.

What's in it, ingredient by ingredient

The useful question is not "does bone stock work" but "what is in it, and what does the evidence say about each of those things." Note that the two columns are independent: an ingredient can be absent from the pot and still have good evidence of its own. MSM has three positive trials and is not in there.

Two rows carry no evidence chip at all, and that is deliberate. Keratan and dermatan sulfate have never been trialled orally in any species, so there is no tier to give them — an em-dash rather than a rating that would imply somebody looked and found nothing.

Where that leaves it

Preclinical, for reasons that are the inverse of what this entry said before. Not "folk tradition plus a contradicted mechanism," but a mechanism whose first links are measured and whose last link has never been tested. Stock delivers gelatin; gelatin yields Pro-Hyp in human blood; Pro-Hyp reaches cartilage in animals and changes what chondrocytes build. Nobody has fed a human bone stock and looked at a joint.

Two honest deflations to keep alongside that. Most of the supporting mechanism work is authored by collagen and gelatin manufacturers. And the composition of any particular pot is close to unknowable — the two properly analysed stocks in the literature differ by roughly 260-fold in glycine content, and nobody has ever analysed one simmered for two days.

None of this makes it bad food. It is safe, pleasant, and cheap if you were making stock anyway, with the caveats that commercial versions carry substantial sodium and histamine-sensitive people anecdotally report trouble with long simmers. Someone who wants a known dose of what the trials tested should take the supplement. Someone who wants to eat well and has a stockpot has no reason to stop. Any human trial at all — even a small biomarker study — would move this entry; to our knowledge, none is registered.