Changelog

2026-10-17 · Capacity audit corrected: 8 of 137 labels, not 36 of 197

On 2026-10-17 the capsule-capacity audit was recomputed: 8 of 137 auditable capsule labels need more compound per capsule than an assumed size 000 capsule holds at 0.8 g/mL, replacing the earlier 36 of 197. Some NIH DSLD records file the salt mass with no form, for example DSLD 297201 files 500 mg for a product that the label sells as 500 mg of glycinate with 70 mg of magnesium; such records are now excluded from the audit with a dated reason. Softgels are no longer checked against hard-capsule capacity, and buffered products and blends no longer receive a single-salt verdict. Elemental fractions were corrected from PubChem: taurate 8.92%, malate dihydrate 12.63%, aspartate up to 15.64% for the 1:1 salt, lactate 10.19% as the dihydrate. Rankings, salt pairs and the calculator read the same shared table. See the label audit.

2026-10-18 · All 716 product-label pages now use one format

The 2026-09-24 test moved half of the product-label pages to the answer-first format. By 2026-10-17, 265 of 358 pages in that format were indexed by Google, against 31 of 358 in the old format. On 2026-10-18 the remaining 358 pages moved to the same format.

2026-09-24 · Pages rebuilt around the answer, with charts and per-page data

The 2026-09-24 releases rebuilt salt pages and within-salt price rankings (including new rankings for malate and taurate) so that each opens with the measured answer, a chart and a page-level CSV. Half of the product-label, salt-pair, brand and facet pages moved to the same format and the other half stay as they were, so the effect of the format can be measured. 104 compact pages were added for ingredients filed on 3 to 9 labels. The price pipeline was repaired: a retail offer now enters a ranking only when brand, salt, form, dose and pack size match the label. The scanned-label link now points to the NIH PDF.

2026-09-13 · Overnight expansion and connected records

The overnight release contains 16 studies and 75 published price observations as of 2026-09-13, alongside new pages derived from the retained source records.

• Methodology and the dataset expose evidence tiers, calculation inputs and dated gaps.
• Excipient facets, filed declarations and capsule shell groups connect source composition fields to their matching labels.
• Formulation clusters compare filed ingredient lists and serving structures across brands; price spreads stay within a single salt.
• Salt pairs show compound mass at equal elemental amounts with hydration intervals.
• The FDA recall register preserves recalling-firm names, product descriptions and source records.
• Food vs supplement compares masses containing 100 mg elemental magnesium.
• 16 absorption studies retain their measured endpoints and populations; 75 published prices retain observation dates and match status.
• Products and brands now link their filed ingredients, declarations and shared formulations; salt pages link their comparisons. Collection entries and breadcrumbs connect these pages to the magnesium hub.

2026-09-11 · Chemistry, measured absorption and prices

The 2026-09-11 revision organized 3 kinds of evidence around each label: salt chemistry, published absorption measurements and dated retail observations.

Salt pages expose elemental fractions and compound mass, the capsule audit shows mass against capacity, and study records attach outcomes to the experiments that measured them. Product pages retain the source of the assigned salt and the basis of each price match.

2026-09-10 · Publication scope narrowed

The 2026-09-10 revision introduced 3 publication tiers: priced, derived and registry-only, separating pages with an observed price or computed quantity from simple registry transcriptions.

Registry-only records remained accessible while their indexing was restricted. Subsequent work added label-specific arithmetic and dated statements of missing information, and reopened those records for indexing. The current editorial policy and methodology describe the published evidence.