Science: research, oceans, ice and how things get measured

Science on this site means the measurement before the conclusion. A study is reported with its method attached (how many subjects, over what period, with what instrument), because a finding without a method is a press release. Where a claim rests on a dataset, the dataset is named and linked, and where the dataset is public the figure on the page is drawn from it rather than quoted from someone else's summary.

The recurring subjects are marine and polar research, climate measurement, species description and the machinery of research itself: peer review, replication, and the gap between what a paper says and what a headline says it says. Several of the pieces here exist because a university or research body cited them, which is the best available signal that the reporting held up.

What this desk avoids is the single-study health story. One trial, one cohort, one surprising correlation, reported as a finding, is the most common failure in science journalism, and the reason is structural rather than malicious: a surprising result is news and a replication is not. Where a study appears here it is placed against what was already known.