We went through the supplied image files at native resolution looking for the Supplement Facts panel, and it is not among them. One product face was supplied — the front — and we transcribed every line on it, including the three claim bars, the strapline, the lung illustration, 2 oz / 60 ml and DIETARY SUPPLEMENT. No photograph of the back or the base was supplied, so we say that rather than describing a panel we have not seen.
We counted the botanicals twice: five named in the formula, ten photographed in the marketing artwork. The five extras are marshmallow root, green tea extract, L-theanine, California poppy seed and corydalis root — two of them sedative rather than respiratory, and one of them the ingredient NCCIH attaches a liver-injury caution to. Both counts are published and the difference is named rather than quietly resolved. We have no reason to think any of the five extras is in the bottle, and with no panel we have no way to confirm that it is not.
We ran the one piece of arithmetic this label permits, twice, in both directions. 60 mL ÷ 0.60 mL is 100 days, which makes the three packs $0.89, $0.59 and $0.49 a day — and makes the six-bottle pack 600 days of product bought against a 60-day refund window. There were no printed percentages to recompute, because there is no panel to print them on.
We opened and read every public source on our reference list, quoted them directly, and included the passages where those sources report their own limitations — the European Medicines Agency's “insufficient evidence from clinical trials” among them. Where we could find no citable source at all, for organic lemon extract, we printed the gap instead of filling it.
And we published the ten supplied reviews in full, with their portraits, labelled as brand-supplied on every card, while emitting no Review and no AggregateRating structured data anywhere, because Google's own guidance is that ratings must be sourced directly from users rather than supplied by the business. We also printed all three of the ratings we were handed — 4.7 claimed, 4.34 recomputed from the distribution, 4.10 across the ten actually published — rather than choosing one. Everything above is checkable by anybody with the same files and an afternoon.