Pull to refresh
Logo
University of California, Berkeley

University of California, Berkeley

University

Appears in 5 stories · Last updated Nov 1, 2026

Stories

Niobium-titanium alloys stay damage-tolerant at 20 kelvin

New Capabilities

Led the research with Lawrence Berkeley National Laboratory and Oxford

At 20 kelvin, just above absolute zero, most body-centered cubic metals turn brittle. Two experimental niobium-titanium alloys did not: they kept absorbing stress without a measurable shift toward brittle fracture.

Updated Nov 1

Pompeii eruption gives scientists a sharper clock for Earth's past

New Capabilities

As of October 2026: Co-author institution

A team of geochronologists used the historically documented eruption of Mount Vesuvius in 79 CE to sharpen argon-argon dating, one of geology's most versatile clocks. The method now dates the eruption to within 13 years of the day Pliny the Younger recorded nearly 2,000 years ago.

Updated Oct 2

Lead in ancient ink makes burned Herculaneum scrolls readable by X-ray

New Capabilities

As of September 2026: Hosted the lead research team

The eruption of Mount Vesuvius in 79 CE turned Herculaneum's papyrus library into brittle black lumps. Physical attempts to unroll the scrolls have destroyed many of them since they were found in 1752. Researchers have now shown a way to read the rest without opening them: lead in the ink.

Updated Sep 26

Genetic study splits gentoo penguins into four species

New Capabilities

As of September 2026: Research and public-communication partner

Gentoo penguins have counted as one species since 1781. A genomic study published in spring 2026 splits them into four, including a previously overlooked species on the remote Kerguelen Islands — the first new penguin named in more than a century.

Updated Sep 22

UC Berkeley researchers release AI model that flags harmful DNA variants with evolutionary history

New Capabilities

As of September 2026: Institution where GPN-Star was developed

UC Berkeley researchers released GPN-Star, an AI model that flags disease-causing DNA variants by learning from millions of years of evolution. Trained in days on eight processors, it outperformed rivals hundreds of times larger at interpreting human DNA.

Updated Sep 20