AstraZeneca reported second-quarter core earnings of $2.63 a share on Monday, up 18% at constant exchange rates, and that is the figure the wires led on. In the same results, filed with the Securities and Exchange Commission as a Form 6-K, reported earnings per share were $1.61 — up 2% in actual currency but down 2% at constant rates. Revenue rose 5% at constant rates to $15,384m, the company reconfirmed its full-year guidance, and it raised the interim dividend three cents to $1.06 a share.

The distance between $2.63 and $1.61 is $1.02, which is 39% of the core figure. Table 11 of the filing sets out the bridge: five cents of restructuring, 79 cents of intangible amortisation and impairments, and 18 cents of other adjustments. At the operating level the gap is wider still — reported operating profit of $3,164m on a 21% margin against core operating profit of $5,158m on a 34% margin, a thirteen-point difference.

Where those add-backs sit is the part that has gone unremarked. Of the $1,546m of amortisation and impairment added back to operating profit, $1,183m was taken out of selling, general and administrative expense and only $355m out of research and development. That single adjustment moves SG&A from 37% of revenue on a reported basis to 26% on a core basis, while R&D barely moves, from 26% to 24%. Excluding acquired-intangible amortisation is a standard and defensible convention in pharmaceuticals; it is worth knowing that in this quarter more than a billion dollars of the cost being set aside was sitting inside the selling line rather than the laboratory.

The tax line did some work too, and the company says so itself. The filing puts the quarter’s reported tax rate at 10%, eleven points below a year earlier, and the core rate at 15%, six points lower, attributing the move to a “benefit from adjustments to deferred tax assets, as a result of certain internal legal entity changes” and noting that “variations in the tax rate can be expected between periods”. In the guidance section of the same document, AstraZeneca states that the core tax rate for 2026 is expected to be between 18% and 22%.

That range implies the quarter was taxed unusually lightly, and the effect on the headline is arithmetically straightforward. Core operating profit of $5,158m less core net finance expense of $340m leaves $4,818m of core pre-tax profit, against which the $731m core tax charge is 15.2%. At the 20% midpoint of the company’s own range the charge would have been roughly $964m — about $233m more, or some fifteen cents spread across the roughly 1.55bn shares implied by the core figures. Core EPS would have come in near $2.48 rather than $2.63, and the growth rate would have been about 11% at constant currency rather than 18%. That is our calculation, holding everything else in the quarter constant, and it is the kind of thing a reconfirmed full-year tax range quietly tells you about the second half.

Underneath the earnings presentation, the business is being carried by one therapy area. Oncology grew 15% at constant rates to $7,327m and now accounts for 48% of revenue. BioPharmaceuticals, the second-largest at 35%, fell 7% to $5,334m, with cardiovascular, renal and metabolism down 18% to $2,772m and infectious disease down 30% to a marginal $131m; respiratory and immunology grew 11%. Rare Disease rose 8% to $2,490m. The company names two causes for the drag: the loss of US exclusivity on Farxiga, and volume-based procurement in China. Chinese revenue fell 13% at constant rates to $1,587m, against US revenue up 6% to $6,686m.

One more line in the revenue table is worth watching. Product sales — the medicines AstraZeneca books at full value — grew 4% at constant rates to $14,510m. Alliance revenue, where it books only its share of a partnered drug, grew 33% to $874m. That is a small base, less than 6% of the total, but it is growing roughly eight times faster than the sales the company keeps whole, which over time changes what a given dollar of headline revenue is worth.

Pascal Soriot, the chief executive, pointed to “six key positive Phase III programmes and eight first approvals in major markets” in the first half, including US clearance for Baxfendy, a first-in-class hypertension medicine. He also acknowledged the setback in CARDIO-TTRansform: “While we are disappointed by the CARDIO-TTRansform outcome, we are on track to deliver our $80bn Total Revenue ambition, which assumes successes and setbacks.” The company says it has more than twenty high-value readouts due over the next eighteen months.

The pattern is becoming familiar. On Sunday we reported that IBM’s segment line reading down 7% concealed mainframe revenue down 42% against distributed infrastructure up 37%, and that Alphabet’s record capital spending of $44.9bn exceeded the quarter’s operating income. AstraZeneca’s version is quieter and entirely conventional: a non-GAAP measure that excludes a real cost, and a tax rate its own guidance says will not persist. The reported number and the core number are both true. They are answers to different questions, and only one of them was in the headlines.