The form is the molecule.

Why the back of a supplement bottle tells you more than the front.

Two bottles on a high-street shelf. Both say "magnesium" in the same point size on the front label. One absorbs at roughly 4% of an ingested dose, measured directly in human plasma. The other absorbs at three to four times that figure, with a side effect of mild glycine intake that turns out to be useful in its own right. The compound on the label is the same word. The thing your body sees is not.

The supplement industry sells you the noun. The mechanism happens at the level of the adjective.

Magnesium oxide is barely a supplement.

If you have ever taken a 400 mg magnesium capsule for sleep or muscle cramps, found that it produced a noticeable laxative effect within a couple of hours, and noticed no change to the symptom you actually bought it for, you have already met magnesium oxide. Magnesium oxide is the most common form sold in the UK, because it carries the largest amount of elemental magnesium per gram on a label and is the cheapest to manufacture. A 400 mg capsule on the back of a Boots multivitamin will, more often than not, be oxide. The number is real. The absorption is not.

Walker and colleagues compared four magnesium forms head to head in healthy adults in 2003, measuring 24-hour urinary excretion and saliva levels. Oxide absorbed at around 10%. Citrate and amino-acid chelates absorbed at around 17%. The clinical literature is consistent: oxide is poorly soluble in the gastric environment, behaves more like a laxative than a systemic mineral, and tends to land at the bottom of every comparative trial.

Magnesium glycinate is the chelated form: magnesium bound to two glycine molecules. The chelation matters for two reasons. First, the bound form is fully soluble at gastric pH and absorbs through amino-acid transport pathways rather than relying on passive diffusion. Second, the glycine released after absorption is itself a neurotransmitter at inhibitory GABA receptors, which gives glycinate its quiet utility for sleep onset that oxide cannot replicate. The same 400 mg of elemental magnesium does not produce the same physiological effect; the form is doing more than the molecule.

The clinical signal lines up. The Kass meta-analysis of 22 magnesium supplementation RCTs in 2012 found a small but real reduction in systolic blood pressure across forms, with effect sizes consistently larger in the trials that used absorbable forms. Abbasi 2012 found magnesium glycinate at 500 mg improved subjective sleep quality and sleep efficiency in adults with insomnia. Neither trial would have produced the same result on equimolar oxide.

A Distil report scoring magnesium as a Foundation-tier recommendation will name glycinate by form, with a target elemental dose, and will explicitly note oxide as the form to avoid. The grade is on the compound. The recommendation is on the form. (If you take it alongside vitamin D, the pairing matters too: see can you take magnesium and vitamin D together?)

Folate and methylfolate are not the same molecule.

If you have a daily multivitamin in a kitchen drawer, the B9 it contains is almost certainly folic acid, and there is roughly a 40% chance your own genetics process that form less efficiently than the version your cells were built to use. Whether you have ever felt the difference depends on what you have been asking folate to do.

Folate is vitamin B9 in the diet: leafy greens, legumes, liver. Folic acid is the synthetic monoglutamate form added to fortified flour and most multivitamins, and it is what gets prescribed to women in pregnancy to prevent neural tube defects. Methylfolate, or L-5-methyltetrahydrofolate, is the form your cells actually use after folate has been processed through the MTHFR enzyme.

About 40% of the European population carries at least one copy of the MTHFR C677T variant, which reduces MTHFR enzyme activity by roughly 30 to 70% depending on heterozygous versus homozygous status. For these people, supplemented folic acid is processed less efficiently into the active methylfolate form, and unmetabolised folic acid can accumulate in serum. The clinical implications are unsettled at the population level, but the biochemistry is not in dispute.

Greenberg 2011 reviewed the methylation pathway and concluded that methylfolate supplementation bypasses the MTHFR bottleneck entirely, delivering the active cofactor directly to one-carbon metabolism. Papakostas 2012 ran a methylfolate augmentation RCT in 75 SSRI-resistant depression patients: methylfolate at 15 mg/day produced a clinically meaningful response advantage over placebo where folic acid in earlier trials had not.

The Distil rule on this is simple. For pregnancy and neural tube prophylaxis, folic acid is the form with the trial evidence and is what NHS guidelines recommend. For mood, methylation, and any context where MTHFR status is a plausible contributor, methylfolate is the form that maps to the mechanism. The compound name "folate" is doing two different jobs depending on what you are asking it to do.

Omega-3 is not omega-3.

If you have a bottle of fish oil on a shelf at home and you bought it for inflammation, joint comfort, or cardiovascular markers, and the result has been hard to feel or hard to measure, the form on the back label is the first variable worth checking. Most fish oil capsules sold in the UK list "EPA" and "DHA" on the back panel without specifying the chemical form. There are three forms in commercial circulation: natural triglyceride, ethyl ester, and re-esterified triglyceride. They are not interchangeable, and the difference matters more than the total milligrams.

Ethyl ester is the cheapest to manufacture and is what most retail fish oil contains. It is also the form with the lowest bioavailability when taken without a high-fat meal. Dyerberg 2010 compared four omega-3 forms in 72 adults over two weeks: the re-esterified triglyceride form produced roughly 24% higher EPA + DHA incorporation into red blood cell membranes than ethyl ester at equal label doses. The natural triglyceride form (the chemical form in actual fish) was intermediate.

This is not a marginal effect. A 1,000 mg EPA dose in ethyl ester form may deliver less absorbed EPA than a 750 mg dose in re-esterified triglyceride form. The label number is not the dose. The form is the dose.

The clinical translation is most visible in cardiovascular trials. REDUCE-IT used pure ethyl-ester icosapent at 4 g/day with food and produced a 25% reduction in major adverse cardiovascular events, but that trial result depended on the high dose, food administration, and a specific high-risk population, not on the form's intrinsic superiority. Lower-dose null trials using ethyl ester without food specification are likely undershooting their nominal dose by a meaningful margin. The Distil report names the form on every omega-3 recommendation, with a strong preference for re-esterified triglyceride and an explicit "take with the largest meal of the day" timing note. The compound is EPA. The recommendation is the form.

Vitamin D2 and D3 do not behave the same.

If you have ever taken vitamin D through a UK winter, had a 25-hydroxyvitamin D blood test afterwards, and watched the number rise less than the dose on the label suggested it should, the form is worth checking before the dose. Cholecalciferol (vitamin D3) is what human skin synthesises from sunlight and what is found in animal-derived food sources. Ergocalciferol (vitamin D2) is what fungi and yeast produce and is what the NHS supplied as the standard supplement form for decades for reasons of cost and shelf stability.

Tripkovic and colleagues published a meta-analysis in 2012 covering ten RCTs comparing D2 and D3 supplementation at equivalent doses. D3 raised serum 25-hydroxyvitamin D roughly twice as effectively as D2 at every dose tested, with the gap widening at higher bolus doses. The mechanism is half-life: D3 binds tightly to vitamin D binding protein and persists in circulation; D2 binds more loosely, is cleared faster, and produces a more transient rise in serum 25(OH)D.

For the rare client on a vegan diet who needs vitamin D supplementation, the choice is D3 from lichen extract over D2 from yeast every time, even though both technically count as "vegan vitamin D." The label words do not capture the bioequivalence gap. The serum measurement does. (We wrote the buyer's version of this up separately: vitamin D2 versus D3, which one to buy.)

Curcumin is bioavailability theatre without context.

If you have ever taken a plain turmeric capsule for joint inflammation, taken it for six weeks, and felt nothing, the bioavailability problem is the most likely explanation. Curcumin is the active compound in turmeric and has Grade B evidence for inflammatory markers, joint pain, and (more cautiously) mood. It also has bioavailability so poor that a 1 g dose of plain curcumin powder produces almost undetectable plasma levels in most adults. The compound is real; the standard form is functionally inert.

Shoba 1998 found that 20 mg of piperine, the active alkaloid in black pepper, increased curcumin bioavailability by approximately 2,000% by inhibiting hepatic glucuronidation. The number is real and is the basis for every "turmeric with black pepper extract" capsule on the market. The trade-off is that piperine also inhibits CYP3A4, the enzyme that metabolises a long list of medications, which moves curcumin from "harmless yellow spice" into "real drug interaction risk" territory. Distil's interactions checker flags curcumin against tacrolimus, ciclosporin, sirolimus, and everolimus as RED hard-exclusions for exactly this reason.

The other route is nanoparticle or lipid encapsulation. Kanai 2012 ran a dose-escalation pharmacokinetic study of Theracurmin, a submicron-particle curcumin formulation, in healthy human volunteers and found plasma curcumin Cmax substantially higher than the published values for plain curcumin powder at equivalent or much larger doses. Meriva and BCM-95 are similar lipid-complexed forms. These do not carry the piperine-CYP3A4 problem and are the form Distil names when curcumin is recommended at all.

The lesson generalises. "Curcumin" on a label tells you almost nothing about whether you are getting a usable systemic dose, a piperine-augmented systemic dose with interaction risk, or essentially food colouring.

The forms also talk to each other.

The five examples do not sit in isolation. A real supplement stack is a network, and the form decisions cascade. Vitamin D3 requires magnesium as a cofactor for both hepatic and renal hydroxylation; supplementing D3 alongside magnesium oxide is buying the active vitamin and an inactive cofactor, and the predictable result is that 25-hydroxyvitamin D fails to rise as expected. Methylfolate and methylcobalamin operate as a methylation pair; one without the other is a one-handed clap, and the trial evidence for either form weakens accordingly when the other is missing. Curcumin with piperine inhibits CYP3A4, the hepatic enzyme through which a long list of medications metabolises, including atorvastatin, simvastatin, and the calcineurin inhibitors used after organ transplant. The form decisions interact. A stack scored at the compound level misses every one of these.

What the back of the bottle should actually say.

The pattern across all five examples is the same. The compound name tells you the rough biological class. The form tells you the dose your body actually sees, the interaction profile, the absorption window, and often the side effect path. A supplement label that names only the compound is failing the same test as a drug label that named only the active pharmaceutical ingredient with no dose, no salt form, and no immediate-release versus extended-release designation. We would not accept that on a prescription bottle. We accept it on a supplement bottle because the regulatory floor is lower and the buyer rarely knows the difference matters.

Every Distil report names the form on every recommendation. Magnesium glycinate, not magnesium. L-methylfolate, not folate. Re-esterified triglyceride EPA, not omega-3. Cholecalciferol, not vitamin D. Lipid-encapsulated curcumin, not turmeric extract. The grade is on the compound. The dose is on the form. The recommendation is on both. (How the grading itself works is in the companion essay, how we grade evidence, and the practical buyer's version is how to read a supplement label.)

The compound name is the cover of the book. The form is the chapter you are actually reading.